®

SAS Programming 3:
Advanced Techniques and
Efficiencies

Course Notes
SAS® Programming 3: Advanced Techniques and Efficiencies Course Notes was developed by Linda
Jolley and Jane Stroupe. Additional contributions were made by Kay Alden, Brian Gayle, Alistair Horn,
Marjorie Lampton, Robert Ligtenberg, Linda Mitterling, Georg Morsing, Kent Reeve, and Jane Whitten.
Editing and production support was provided by the Curriculum Development and Support Department.
SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of
SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product
names are trademarks of their respective companies.

SAS® Programming 3: Advanced Techniques and Efficiencies Course Notes

Copyright © 2010 SAS Institute Inc. Cary, NC, USA. All rights reserved. Printed in the United States of
America. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in
any form or by any means, electronic, mechanical, photocopying, or otherwise, without the prior written
permission of the publisher, SAS Institute Inc.

Book code E1833, course code LWPRG3/PRG3, prepared date 16Sep2010. LWPRG3_003

ISBN 978-1-60764-748-5
 For Your Information iii

Table of Contents

Course Description ...................................................................................................................... ix

Prerequisites ................................................................................................................................. x

Chapter 1 Introduction .......................................................................................... 1-1

1.1 Course Logistics .............................................................................................................. 1-3

1.2 Measuring Efficiencies .................................................................................................... 1-9

1.3 SAS DATA Step Processing ........................................................................................... 1-29

Exercises.................................................................................................................. 1-33

1.4 Chapter Review.............................................................................................................. 1-36

1.5 Solutions ........................................................................................................................ 1-37

Solutions to Exercises ............................................................................................. 1-37
Solutions to Student Activities (Polls/Quizzes) ....................................................... 1-40
Solutions to Chapter Review ................................................................................... 1-42

Chapter 2 Controlling I/O Processing and Memory ............................................ 2-1

2.1 Controlling I/O ................................................................................................................. 2-3

Exercises.................................................................................................................. 2-13

2.2 Controlling Data Set Size............................................................................................... 2-16

Exercises.................................................................................................................. 2-26

2.3 Compressing SAS Data Sets .......................................................................................... 2-29

Exercises.................................................................................................................. 2-43

2.4 Controlling Memory (Self-Study).................................................................................. 2-45

2.5 Controlling the Page Size and the Number of Available Buffers (Self-Study) .............. 2-53

2.6 Chapter Review.............................................................................................................. 2-59

2.7 Solutions ........................................................................................................................ 2-60

iv For Your Information

Solutions to Exercises ............................................................................................. 2-60

Solutions to Student Activities (Polls/Quizzes) ....................................................... 2-67
Solutions to Chapter Review ................................................................................... 2-71

Chapter 3 Accessing Observations...................................................................... 3-1

3.1 Creating an Index ............................................................................................................. 3-3

Exercises.................................................................................................................. 3-28

3.2 Using an Index ............................................................................................................... 3-29

Exercises.................................................................................................................. 3-47

3.3 Creating a Sample Data Set (Self-Study) ....................................................................... 3-50

Exercises.................................................................................................................. 3-70

3.4 Chapter Review.............................................................................................................. 3-71

3.5 Solutions ........................................................................................................................ 3-72

Solutions to Exercises ............................................................................................. 3-72
Solutions to Student Activities (Polls/Quizzes) ....................................................... 3-80
Solutions to Chapter Review ................................................................................... 3-86

Chapter 4 Introduction to Lookup Techniques ................................................... 4-1

4.1 Introduction to Lookup Techniques ................................................................................. 4-3

4.2 In-Memory Lookup Techniques....................................................................................... 4-5

4.3 Disk Storage Techniques ................................................................................................ 4-13

4.4 Chapter Review.............................................................................................................. 4-28

4.5 Solutions ........................................................................................................................ 4-29

Solutions to Student Activities (Polls/Quizzes) ....................................................... 4-29
Solutions to Chapter Review ................................................................................... 4-32

Chapter 5 Using DATA Step Arrays ...................................................................... 5-1

5.1 Using One-Dimensional Arrays ....................................................................................... 5-3

 For Your Information v

Exercises.................................................................................................................. 5-18

5.2 Using Multidimensional Arrays ..................................................................................... 5-22

Exercises.................................................................................................................. 5-35

5.3 Loading a Multidimensional Array from a SAS Data Set.............................................. 5-40

Exercises.................................................................................................................. 5-63

5.4 Chapter Review.............................................................................................................. 5-69

5.5 Solutions ........................................................................................................................ 5-70

Solutions to Exercises ............................................................................................. 5-70
Solutions to Student Activities (Polls/Quizzes) ....................................................... 5-79
Solutions to Chapter Review ................................................................................... 5-83

Chapter 6 Using DATA Step Hash and Hiter Objects .......................................... 6-1

6.1 Introduction ...................................................................................................................... 6-3

6.2 Using Hash Object Methods ............................................................................................ 6-7

Exercises.................................................................................................................. 6-28

6.3 Loading a Hash Object with Data from a SAS Data Set................................................ 6-31
Exercises.................................................................................................................. 6-42

6.4 Using the DATA Step Hiter Object ................................................................................ 6-48

Exercises.................................................................................................................. 6-65

6.5 Using a Hash Object for Chained Lookups (Self-Study) ............................................... 6-67
Demonstration: Creating a List of Values ............................................................... 6-82
Exercises.................................................................................................................. 6-84

6.6 Chapter Review.............................................................................................................. 6-87

6.7 Solutions ........................................................................................................................ 6-89

Solutions to Exercises ............................................................................................. 6-89
Solutions to Student Activities (Polls/Quizzes) ..................................................... 6-100
Solutions to Chapter Review ................................................................................. 6-105
vi For Your Information

Chapter 7 Creating and Using Formats ............................................................... 7-1

7.1 Using Formats as Lookup Tables ..................................................................................... 7-3

Demonstration: Using a Control Data Set to Create a Format .................................. 7-5
Exercises.................................................................................................................. 7-21

7.2 Using a Picture Format (Self-Study) .............................................................................. 7-24

Exercises.................................................................................................................. 7-34

7.3 Chapter Review.............................................................................................................. 7-36

7.4 Solutions ........................................................................................................................ 7-37

Solutions to Exercises ............................................................................................. 7-37
Solutions to Student Activities (Polls/Quizzes) ....................................................... 7-42
Solutions to Chapter Review ................................................................................... 7-46

Chapter 8 Combining Data Horizontally............................................................... 8-1

8.1 DATA Step Merges and SQL Procedure Joins ................................................................. 8-3
Demonstration: Using the DATA Step to Perform a Match-Merge........................... 8-7
Demonstration: Using a PROC SQL Join to Perform a Match-Merge ................... 8-10
Exercises (Optional) ................................................................................................ 8-17

8.2 Using an Index to Combine Data ................................................................................... 8-22

Demonstration: Using Multiple SET … KEY= Statements (Self-Study) ............... 8-40
Exercises.................................................................................................................. 8-42

8.3 Combining Summary and Detail Data ........................................................................... 8-49

Exercises.................................................................................................................. 8-64

8.4 Combining Data Conditionally (Self-Study) ................................................................. 8-71

Demonstration: Using a Hash Object ...................................................................... 8-90
Exercises.................................................................................................................. 8-92

8.5 Chapter Review.............................................................................................................. 8-96

8.6 Solutions ........................................................................................................................ 8-98

 For Your Information vii

Solutions to Exercises ............................................................................................. 8-98

Solutions to Student Activities (Polls/Quizzes) ..................................................... 8-116
Solutions to Chapter Review ................................................................................. 8-123

Chapter 9 Sorting SAS Data Sets ......................................................................... 9-1

9.1 Using the SORT Procedure .............................................................................................. 9-3

Demonstration: Using the EQUALS and NOEQUALS Options ............................ 9-21
Demonstration: Using the NUMERIC_COLLATION= Option ............................. 9-27
Exercises.................................................................................................................. 9-30

9.2 BY-Group Processing (Self-Study) ................................................................................ 9-33

Exercises.................................................................................................................. 9-53

9.3 Chapter Review.............................................................................................................. 9-58

9.4 Solutions ........................................................................................................................ 9-59

Solutions to Exercises ............................................................................................. 9-59
Solutions to Student Activities (Polls/Quizzes) ....................................................... 9-66
Solutions to Chapter Review ................................................................................... 9-74

Chapter 10 Programmer Efficiency ...................................................................... 10-1

10.1 Introduction .................................................................................................................... 10-3

10.2 Writing Flexible Programs: Combining Raw Data Files Vertically ............................. 10-10
Exercises................................................................................................................ 10-46

10.3 Creating Views ............................................................................................................. 10-51

Demonstration: Creating a DATA Step View ........................................................ 10-54
Exercises................................................................................................................ 10-74

10.4 Using FILE and PUT Statements to Create a SAS Program File ................................ 10-78
Demonstration: Using the DATA Step to Send E-Mail ......................................... 10-87
Exercises................................................................................................................ 10-90

10.5 Using the FCMP Procedure (Self-Study) ..................................................................... 10-95

viii For Your Information

Demonstration: Creating and Using Functions ................................................... 10-106

Demonstration: Creating Subroutines Using PROC FCMP ............................... 10-110
Exercises.............................................................................................................. 10-115

10.6 Chapter Review.......................................................................................................... 10-119

10.7 Solutions .................................................................................................................... 10-121

Solutions to Exercises ......................................................................................... 10-121
Solutions to Student Activities (Polls/Quizzes) ................................................... 10-134
Solutions to Chapter Review ............................................................................... 10-140

Chapter 11 Customizing Your SAS Session (Self-Study) ................................... 11-1

11.1 Introduction .................................................................................................................... 11-3

11.2 Editing the Configuration File ....................................................................................... 11-7

11.3 Creating an Autoexec.sas File ...................................................................................... 11-22

11.4 Using the SAS Registry ............................................................................................... 11-28

11.5 Solutions ...................................................................................................................... 11-40

Solutions to Student Activities (Polls/Quizzes) ..................................................... 11-40

Chapter 12 Learning More ..................................................................................... 12-1

12.1 Conclusions .................................................................................................................... 12-3

12.2 SAS Resources ............................................................................................................... 12-8

12.3 Beyond This Course ..................................................................................................... 12-12

Appendix A Index ..................................................................................................... A-1

 For Your Information ix

Course Description
• This course is for SAS programmers who prepare data for analysis. The comparisons of manipulation
techniques and resource cost benefits are designed to help programmers choose the most appropriate
technique for their data situation.You will learn how to compare various SAS programming techniques
that enable you to
• control memory, I/O, and CPU resources
• create and use indexes
• combine data horizontally and vertically
• use hash and hiter DATA step component objects, arrays, and formats as lookup tables
• compress SAS data sets
• sample your SAS data sets.

To learn more…

For information on other courses in the curriculum, contact the SAS Education
Division at 1-800-333-7660, or send e-mail to [email protected]. You can also
find this information on the Web at support.sas.com/training/ as well as in the
Training Course Catalog.

For a list of other SAS books that relate to the topics covered in this
Course Notes, USA customers can contact our SAS Publishing Department at
1-800-727-3228 or send e-mail to [email protected]. Customers outside the
USA, please contact your local SAS office.
Also, see the Publications Catalog on the Web at support.sas.com/pubs for a
complete list of books and a convenient order form.
x For Your Information

Prerequisites
This course is not appropriate for beginning SAS software users. Before attending this course, you should
have at least nine months of SAS programming experience and should have completed the SAS®
Programming 2: Data Manipulation Techniques course. Specifically, you should be able to do the
following:
• understand your operating system file structures and perform basic operating system tasks
• understand programming logic concepts
• understand the compilation and execution process of the DATA step
• use different varieties of input to create SAS data sets from external files
• use SAS software to access SAS libraries
• create and use SAS date values
• read, concatenate, merge, match-merge, and interleave SAS data sets
• use the DROP=, KEEP=, and RENAME= data set options
• create multiple output data sets
• use array processing and DO loops to process data iteratively
• use SAS functions to perform data manipulation and transformations
Chapter 1 Introduction

1.1 Course Logistics ............................................................................................................ 1-3

1.2 Measuring Efficiencies................................................................................................... 1-9

1.3 SAS DATA Step Processing ......................................................................................... 1-29

Exercises .............................................................................................................................. 1-33

1.4 Chapter Review............................................................................................................. 1-36

1.5 Solutions ....................................................................................................................... 1-37

Solutions to Exercises .......................................................................................................... 1-37

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 1-40

Solutions to Chapter Review ................................................................................................ 1-42

1-2 Chapter 1 Introduction
 1.1 Course Logistics 1-3

1.1 Course Logistics

Objectives
„ List the tasks in the SAS Programming 3 course.
„ Explain the naming convention that is used for the
course files.
„ Compare the three levels of exercises that are used
in the course.
„ Describe, at a high level, how data is used and stored
at Orion Star Sports & Outdoors.
„ Navigate to the Help facility.

Tasks in the SAS Programming 3 Course

The course topics include techniques for the following
data management tasks:
„ compressing SAS data sets

„ creating indexes for a quick retrieval of subsets

„ performing table lookups using arrays, hash objects,

or formats
„ combining data by merging, using the SQL procedure,
or using multiple SET statements
„ combining summary and detail data

„ sorting and grouping data

„ developing a program quickly

4
1-4 Chapter 1 Introduction

Resource Utilization
As programmers, you want to perform these tasks
as efficiently as possible and optimize the use of the
following resources:
„ programmer time

„ I/O

„ CPU

„ memory

„ data storage space

„ network bandwidth

Business Scenarios
The business scenarios are opportunities to compare
multiple techniques for performing the tasks.
For example:
„ Task: Table Lookups

„ Possible Techniques:

– DATA step MERGE statement

– PROC SQL joins
– Formats in PUT functions or in FORMAT
statements
– DATA step arrays
– DATA step hash objects

6
 1.1 Course Logistics 1-5

1.01 Multiple Answer Poll

What type(s) of SAS programs do you write?
a. Data manipulation with the DATA step
b. Data analysis with procedures
c. Report writing
d. A combination of the above
e. SAS training only; no programs written
f. Other

Filename Conventions
p304d01x

course ID chapter # type item # placeholder

p304a01
Code Type p304a02 Example:
p304a02s The SAS Programming 3
a Activity
course ID is p3, so
p304d01
d Demo p304d01 =
p304d02 SAS Programming 3,
e Exercise p304e01 Chapter 4, Demo 1.
s Solution p304e02
p304s01
p304s02
9
1-6 Chapter 1 Introduction

Three Levels of Exercises

Level 1 The exercise mimics an example
presented in the section.

Level 2 Less information and guidance are

provided in the exercise instructions.
Level 3 Only the task you are to perform or
the results to be obtained are provided.
Typically, you will need to use the
Help facility.

 You are not expected to complete all of the exercises

in the time allotted. Choose the exercise or exercises
that are at the level with which you are most
comfortable.
10

Orion Star Sports & Outdoors

Orion Star Sports & Outdoors is a fictitious global sports

and outdoors retailer with traditional stores, an online store,
and a large catalog business.

The corporate headquarters is located in the United States

with offices and stores in many countries throughout the
world.

Orion Star has about 1,000 employees and 90,000

customers, processes approximately 150,000 orders
annually, and purchases products from 64 suppliers.

11
 1.1 Course Logistics 1-7

Orion Star Data

As is the case with most organizations, Orion Star has
a large amount of data about its customers, suppliers,
products, and employees. Much of this information is
stored in transactional systems in various formats.

Using applications and processes such as SAS Data

Integration Studio, this transactional information was
extracted, transformed, and loaded into a data
warehouse.

Data marts were created to meet the needs of specific

departments such as Marketing.

12

The SAS Help Facility

13
1-8 Chapter 1 Introduction

1.02 Quiz
„ Start your SAS session.
„ Open the Help facility.
„ Determine the path to use to obtain information about
the SAS component objects.

15

SAS OnlineDoc
You can also obtain information from SAS OnlineDoc.

Information relevant to this

course can be found by
following these paths in
SAS OnlineDoc:
Contents tab
Æ Products Documentation
A-Z
Æ Base SAS
Æ SAS 9.2 Language
Reference Dictionary
Æ Dictionary of
Component
Object Language
Elements

17
 1.2 Measuring Efficiencies 1-9

1.2 Measuring Efficiencies

Objectives
„ Identify the resources used by a SAS program.
„ Report computer resource usage using SAS system
options.
„ Interpret resource usage statistics in your operating
environment.
„ Benchmark resource usage.

20

Running a SAS Program

What resources are required to run a SAS program?
The programmer must perform the following tasks:
„ determine program specifications

„ write the program

„ test the program

„ execute the program

„ maintain the program

21
1-10 Chapter 1 Introduction

Running a SAS Program

The computer must perform the following actions:
„ load the required SAS software into memory

„ compile the program

„ read the data

„ execute the compiled program

„ store output data files

„ store output reports

22
 1.2 Measuring Efficiencies 1-11

What Resources Are Used?

CPU
programmer
I/O
time

resources used

network memory
bandwidth

data storage
space

23

CPU is a measurement of the amount of time that the central processing unit
uses to perform requested tasks such as calculations, reading and writing
data, conditional and iterative logic, and so on.

I/O is a measurement of the read-and-write operations performed when data

and programs are moved from a storage device to memory (input) or from
memory to a storage or display device (output).

Memory is the size of the work area required to hold executable program modules,
data, and buffers.

Data storage space is the amount of space that is required to store data on a disk or tape.

Programmer time is the amount of time required for the programmer to write and maintain
the program. This can be decreased through well-documented, best
programming practices.

Network bandwidth is the amount of data that can pass through a network interface over time.
This time can be minimized by performing as much of the subsetting and
summarizing as possible on the data host before transferring the results to
the local computer. The network bandwidth is heavily dependent on
network loads.
1-12 Chapter 1 Introduction

1.03 Multiple Answer Poll

Which of the following resources do you need to
conserve?
a. CPU
b. I/O
c. Memory
d. Data storage space
e. Network bandwidth
f. Your time

25

Understanding Efficiency Trade-offs

When you decrease the use of one resource, the use
of other resources might increase.
Resource usage is dependent on your data. A specific
technique might be more efficient with one data set and
less efficient with another.

26
 1.2 Measuring Efficiencies 1-13

Understanding Efficiency Trade-offs

Data Data

Space

Decreasing the size 12

of a SAS data set can 12
9 3
result in an increase in 9 3

6
CPU usage. 6

CPU

27 ...

For example, data file compression might decrease storage use but increase processing time when SAS
reads the compressed data.

Understanding Efficiency Trade-offs

I/O

Decreasing the number

of I/O operations comes
at the expense of increased
memory usage.
Memory

28

I/O can be decreased by increasing buffering space and memory usage.

1-14 Chapter 1 Introduction

Deciding What Is Important for Efficiency

Your Programs

Your Site Your Data

29

You must decide which factors are the most important for improving resource usage at your site.
To make this decision, you must know the following:
• which resources are scarce or costly at your site
• how and when your programs will be used
• the type and volume of data that your programs will process
 1.2 Measuring Efficiencies 1-15

Understanding Efficiency at Your Site

Hardware Operating Environment

System Load
SAS Environment

30

Environmental factors that affect the efficiency of SAS programs include the following:

Hardware number of CPUs, amount of available memory, number of I/O

controllers for data access, network infrastructure

Operating environment resource allocation, scheduling algorithms, and I/O methods

System load the number of users or jobs sharing system resources, including
network bandwidth and network traffic

SAS environment which SAS software products are installed, how they were installed,
and which methods are available to run SAS programs at your site
Often, one or two resources constitute the limiting factor or bottleneck within an organizational
computing environment. Tuning can be used to shift dependence away from a constrained resource. By
tuning your SAS programs to use the more available resources, you might improve the performance.
1-16 Chapter 1 Introduction

1.04 Multiple Choice Poll

This class uses SAS 9.2.
What is the latest version of SAS that are you running?
a. SAS 8.2
b. SAS 9.1
c. SAS 9.2
d. Other

32

Knowing How Your Program Will Be Used

The importance of efficiency increases with the following:
„ the complexity of the program and/or the size of the
files being processed
„ the number of times that the program will be executed

33
 1.2 Measuring Efficiencies 1-17

Knowing Your Data

34

When you know the characteristics of your data, you can select the techniques that best suit those
characteristics.

1.05 Multiple Answer Poll

What type(s) of data do you use?
a. SAS data sets
b. External files
c. Data from a relational database – for example,
Oracle, Teradata, or SQL Server
d. Excel spreadsheets
e. OLAP cubes
f. Information maps
g. Other

36
1-18 Chapter 1 Introduction

Considering Trade-Offs
In this class, many tasks are performed using one or more
techniques.
To decide which technique is most efficient for a given
task, benchmark, or measure and compare, the resource
usage of each technique.
You should benchmark with the actual data to determine
which technique is the most efficient.

The effectiveness of any efficiency technique

depends greatly on the data with which you use
the technique.

37

Running Benchmarks: Guidelines

To benchmark your programming techniques, do the
following:
„ Turn on the appropriate options to report resource
usage.
„ Test each technique in a separate SAS session.

„ Test only one technique or change at a time, with

as little additional code as possible.
„ Run your tests under the conditions that your final
program will use (for example, batch execution,
large data sets, and so on).

38 continued...
 1.2 Measuring Efficiencies 1-19

Running Benchmarks: Guidelines

„ Run each program several times and base your
conclusions on averages, not on a single execution.
(This is more critical when you benchmark elapsed
time.)
„ Exclude outliers from the analysis because that data
might lead you to tune your program to run less
efficiently than it should.
„ Turn off the options that report resource usage after
testing is finished, because they consume resources.

In a multi-user environment, other computer

activities might affect the running of your program.

39

1.06 Multiple Choice Poll

Which of the following SAS programs should be
benchmarked?
a. A report that shows all the customers in the United
Kingdom in March 2006
b. A report that calculates trends in sales at the end
of every day for every department
c. A report showing the projected total cost of a 5%
cost-of-living increase in employee salaries for a
Human Resources project conducted on January 1,
2007
d. A yearly report that calculates the average sales
of a line of apparel for the clothing manager

41
1-20 Chapter 1 Introduction

Tracking Resource Usage

STIMER

STATS SAS MEMRPT

(z/OS only) Options (z/OS only)

FULLSTIMER

43

There are four SAS system options that you can use to track and report on resource utilization:

STIMER tracks the CPU time used to perform a task (DATA or PROC step).

MEMRPT tracks memory used to perform a task. z/OS only

FULLSTIMER tracks usage of additional resources and divides CPU into system CPU
time and user CPU time. This option is ignored unless STIMER or
MEMRPT is in effect.

STATS writes information tracked by the above options to the SAS log. z/OS only

 The availability, usage, and aliases of these options are specific to the operating environment.
 1.2 Measuring Efficiencies 1-21

Tracking Resources with SAS Options

Windows, UNIX

OPTIONS STIMER | NOSTIMER;

OPTIONS NOFULLSTIMER | FULLSTIMER;

z/OS
STIMER» | NOSTIMER Invocation option only

OPTIONS STATS | NOSTATS;

OPTIONS MEMRPT | NOMEMRPT;

OPTIONS NOFULLSTIMER | FULLSTIMER;

44

z/OS Windows UNIX

FULLSTIMER B B B

STIMER ID BD BD

STATS BD N/A N/A

MEMRPT BD N/A N/A

I Invocation option only

B Can be set at invocation or by using an OPTIONS statement
N/A Not available (The functionality is part of the STIMER option under UNIX and Windows.)
D Default
You can find more information about operating environment dependencies in the SAS documentation for
your operating environment.

 Use the OPTIONS procedure with the HOST option to determine the default settings of these
options at your site.
proc options host;
run;
1-22 Chapter 1 Introduction

Business Scenario
You should benchmark to determine the most efficient
technique for creating a new variable based on a
condition.
The following methods can be used:
„ IF-THEN with an assignment statement

„ IF-THEN/ELSE with an assignment statement

„ SELECT/WHEN with an assignment statement

45

1.07 Quiz
1. Open and submit p301a01a.
Record the user CPU: ____________
Exit SAS.
2. Start SAS.
Open and submit p301a01b.
Record the user CPU: ____________
Exit SAS.
3. Start SAS.
Open and submit p301a01c.
Record the user CPU: ____________
4. Which technique is most efficient?
In z/OS, record the CPU.
47
 1.2 Measuring Efficiencies 1-23

p301a01a
options fullstimer;
data _null_;
length var $ 30;
retain var2-var50 0 var51-var100 'ABC';
do x=1 to 10000000;
var1=10000000*ranuni(x);
if var1>1000000 then var='Greater than 1,000,000';
if 500000<=var1<=1000000
then var='Between 500,000 and 1,000,000';
if 100000<=var1<500000 then var='Between 100,000 and 500,000';
if 10000<=var1<100000 then var='Between 10,000 and 100,000';
if 1000<=var1<10000 then var='Between 1,000 and 10,000';
if var1<1000 then var='Less than 1,000';
end;
run;
p301a01b
options fullstimer;
data _null_;
length var $ 30;
retain var2-var50 0 var51-var100 'ABC';
do x=1 to 10000000;
var1=10000000*ranuni(x);
if var1>1000000 then var='Greater than 1,000,000';
else if 500000<=var1<=1000000
then var='Between 500,000 and 1,000,000';
else if 100000<=var1<500000
then var='Between 100,000 and 500,000';
else if 10000<= var1<100000
then var='Between 10,000 and 100,000';
else if 1000<=var1<10000 then var='Between 1,000 and 10,000';
else if var1<1000 then var='Less than 1,000';
end;
run;
(Continued on the next page.)
1-24 Chapter 1 Introduction

p301a01c
options fullstimer;
data _null_;
length var $ 30;
retain var2-var50 0 var51-var100 'ABC';
do x=1 to 10000000;
var1=10000000*ranuni(x);
select;
when (var1>1000000) var='Greater than 1,000,000';
when (500000<=var1<=1000000)
var='Between 500,000 and 1,000,000';
when (100000<=var1<500000) var='Between 100,000 and 500,000';
when (10000<=var1<100000) var='Between 10,000 and 100,000';
when (1000<=var1<10000) var='Between 1,000 and 10,000';
when (var1<1000) var='Less than 1,000';
otherwise;
end;
end;
run;
 1.2 Measuring Efficiencies 1-25

Sample Windows Log

Partial SAS Log
5 options fullstimer;
6 data _null_;
7 length var $ 30;
8 retain var2-var50 0 var51-var100 'ABC';
9 do x=1 to 100000000;
10 var1=10000000*ranuni(x);
11 if var1>1000000 then var='Greater than 1,000,000';
12 if 500000<=var1<=1000000 then var='Between 500,000 and 1,000,000';
13 if 100000<=var1<500000 then var='Between 100,000 and 500,000';
14 if 10000<=var1<100000 then var='Between 10,000 and 100,000';
15 if 1000<=var1<10000 then var='Between 1,000 and 10,000';
16 if var1<1000 then var='Less than 1,000';
17 end;
18 run;
NOTE: DATA statement used (Total process time):
real time 1.26 seconds
user cpu time 0.98 seconds
system cpu time 0.04 seconds
Memory 278k
OS Memory 4976k
Timestamp 6/29/2010 12:39:21 PM

48 p301a01a

Description of FULLSTIMER statistics in the Windows operating environment:

Real Time the amount of time spent to process the SAS job. (Real time is also referred
to as elapsed time.)

User CPU Time the CPU time spent to execute the SAS code as written by the user

System CPU Time the CPU time spent to perform operating system tasks (system overhead
tasks) that support the execution of SAS code

Memory the amount of memory required to run a step

OS Memory the largest amount of memory that SAS requested from the operating
system during the step

Timestamp the date and time that the statistics were produced
1-26 Chapter 1 Introduction

Sample UNIX Log

Partial SAS Log
1 options fullstimer;
2 data _null_;
3 length var $30;
4 retain var2-var50 0 var51-var100 'ABC';
5 do x=1 to 10000000;
6 var1=10000000*ranuni(x);
7 if var1>10000000 then var='Greater than 1,000,000';
8 if 500000<=var1<=1000000 then var='Between 500,000 and 1,000,000';
9 if 100000<=var1<500000 then var='Between 100,000 and 500,000';
10 if 10000<=var1<100000 then var='Between 10,000 and 100,000';
11 if 1000<=var1<10000 then var='Between 1,000 and 10,000';
12 if var1<1000 then var='Less than 1,000';
13 end;
14 run;

NOTE: DATA statement used (Total process time):

real time 6.62 seconds
user cpu time 5.14 seconds
system cpu time 0.01 seconds
Memory 526k
OS Memory 5680k
Timestamp 6/29/2010 11:55:32 AM
Page Faults 82
Page Reclaims 0
Page Swaps 0
Voluntary Context Switches 91
Involuntary Context Switches 48
Block Input Operations 91
Block Output Operations 0

49 p301a01a

SAS uses the getrusage() and times() UNIX system calls for your operating environment to obtain the
statistics presented with FULLSTIMER.

 Different “flavors” of UNIX show different statistics. This log was obtained on a Solaris system.
Description of FULLSTIMER statistics in the UNIX operating environment:

Real Time the amount of time spent to process the SAS job. (Real time is also referred
to as elapsed time.)

User CPU Time the CPU time spent to execute the SAS code as written by the user

System CPU Time the CPU time spent to perform operating system tasks (system overhead
tasks) that support the execution of SAS code

Memory the amount of memory required to run a step

OS Memory the largest amount of memory that SAS requested from the operating
system during the step

Timestamp the date and time that the statistics were produced

Page Faults the number of pages that SAS tried to access but were not in the main
memory and required I/O activity

Page Reclaims the number of pages that were accessed without I/O activity
(Continued on the next page.)
 1.2 Measuring Efficiencies 1-27

Page Swaps the number of times that a SAS process was swapped out of main memory

Voluntary Context the number of times that the SAS process had to pause because of a
Switches resource constraint such as a disk drive

Involuntary the number of times that the operating system forced the SAS session to
Context Switches pause processing to enable other processes to run

Block Input the number of I/O operations that were performed to read the data into
Operations memory

Block Output the number of I/O operations that were performed to write the data to a file
Operations
1-28 Chapter 1 Introduction

Sample z/OS Log

Partial SAS Log

50 p301a01a

Description of FULLSTIMER statistics in the z/OS operating environment:

CPU Time The actual time spent on the task. This number should be constant (within
.02 seconds) across repetitions of the same job.

Elapsed Time The wall-clock time required to complete the task. Because elapsed time
varies greatly for several runs of the same job due to differences in waiting
time caused by other tasks being performed by the CPU, it is not normally
used to benchmark programs.

EXCP Count The number of I/O operations required to transfer external data to and from
memory. EXCP is the acronym for EXecute Channel Program.

Task Memory The actual memory required for a task in kilobytes with breakdowns for
data and program storage. This number is stable for a given task.

Total Memory The memory required for all tasks in kilobytes. This session total is useful
for deciding the minimum region size required for the job to execute
successfully.
 1.3 SAS DATA Step Processing 1-29

1.3 SAS DATA Step Processing

Objectives
„ List the attributes of a data set page and define how
it relates to the structure of SAS data sets.
„ Describe how SAS reads and writes data.

53

SAS Data Set Pages

A SAS data set page has the following attributes:
„ It is the unit of data transfer between the operating
system buffers and SAS buffers in memory.
„ It includes the number of bytes used by the descriptor
portion, the data values, and any operating system
overhead.
„ It is fixed in size when the data set is created, either
to a default value or to a value specified by the
programmer.

54

By default, SAS uses the minimum optimal page size for the operating environment.
1-30 Chapter 1 Introduction

Using PROC CONTENTS to Report Page Size

proc contents data=orion.sales_history;
run;

Partial PROC CONTENTS Output 16,384*18=

294,912 bytes
Engine/Host Dependent Information

Data Set Page Size 16384

Number of Data Set Pages 18
First Data Page 1
Max Obs per Page 92
Obs in First Data Page 72
Number of Data Set Repairs 0
File Name S:\workshop\sales_history.sas7bdat
Release Created 9.0201M0
Host Created XP_PRO

55

The total number of bytes occupied by orion.sales_history can be calculated as shown below:
(16,384 * 18)=294,912 bytes

1.08 Quiz
Use one of the following to determine the page size
of the orion.customer_dim SAS data set:
„ the CONTENTS procedure

„ the DATASETS procedure

„ the SAS Explorer window

What is the page size of the SAS data set

orion.customer_dim?

p301a02
57
 1.3 SAS DATA Step Processing 1-31

Reading External Files

I/O
measured Input Buffer
Input here
Raw Buffers
Data
Data is converted
from external
Caches memory format to
SAS format.
PDV
Output ID Gender Country Name

SAS I/O Buffers

Data measured
here

64

When a raw data file is read with INFILE and INPUT statements, the following occur:
• A block of data is read into a buffer in memory. The size of each buffer is the block size of the input
raw data file. In Windows and UNIX, the data might be cached so that the data is copied from disk to
an area of memory managed by the operating environment before it is copied into the buffer managed
by SAS.
• Each record of the raw data file is copied into an input buffer.
• The data is converted from an external format to the SAS format using the instructions provided in the
INPUT statement and is stored in an area of SAS memory named the program data vector (PDV). Any
subsequent processing specified in the DATA step is performed on the values in the PDV.
• At the end of an iteration for the DATA step, the contents of the PDV are copied to an output buffer
in memory.
• After the buffer (or multiple buffers) is full, the data in the buffer is written to the output SAS data set
in one output operation.
• Sequential processing continues until the pointer reaches the end-of-file marker in the raw data file.
1-32 Chapter 1 Introduction

Reading a SAS Data Set with a SET Statement

I/O
measured
Input here No data
SAS conversion
Data is necessary.

Caches memory

PDV
ID Gender Country Name
Output
SAS I/O
Data measured
here

71 ...

When a SAS data set is read with a SET or MERGE statement, the following occur:
• A page (or multiple pages) is read into a buffer (or multiple buffers) in memory. The size of each buffer
is the page size of the input SAS data set. In Windows and UNIX, the data might be cached so that the
data is copied from disk to an area of memory managed by the operating environment before it is
copied into the buffer managed by SAS.
• The data in the buffer is read sequentially and copied into the program data vector (PDV) one
observation at a time.
• Each observation of the new SAS data set is copied into a buffer (or multiple buffers). An observation
must fit entirely into the buffer or the observation is written to another buffer.
• After the buffer (or multiple buffers) is full, the data in the buffer is written to the output SAS data set
in one output operation.
• Sequential processing continues until the pointer reaches the end-of-file marker in the input SAS data
set.
 1.3 SAS DATA Step Processing 1-33

Exercises

Level 1

1. Benchmarking
Open the program p301e01.sas (Windows or UNIX) or '.prg3.sascode(p301e01)' (z/OS).
Use best practices to benchmark the program, change it according to step 1.d, and determine which
resource(s) were conserved.
data order_fact;
infile 'order_fact.dat' pad; M
input @37 Order_Date date9. @; N
input @1 Customer_ID 12.
@13 Employee_ID 12.
@25 Street_ID 12.
@46 Delivery_Date date9.
@55 Order_ID 12.
@67 Order_Type 2.
@69 Product_ID 12.
@81 Quantity 4.
@90 Total_Retail_Price 13.
@105 CostPrice_Per_Unit 10.
@115 Discount 5.;
if year(Order_Date)=2006;
format Customer_ID Employee_ID Street_ID Order_ID
Product_ID 12. Order_Date Delivery_Date date9.
Order_Type 2. Quantity 4. Total_Retail_Price dollar13.2
CostPrice_Per_Unit dollar10.2 Discount Percent.;
run;
Notes about the syntax:

c PAD controls whether SAS pads the records that are read from an external file with
blanks to the length that is specified in the LRECL= option. The LRECL=
option specifies the logical record length; it is dependent on the operating
environment.

d @ holds an input record for the execution of the next INPUT statement within the
same iteration of the DATA step. This line-hold specifier is called a trailing @.
a. Turn on the appropriate options for reporting the resource statistics in the log.
b. Submit the program.
1-34 Chapter 1 Introduction

c. Record the following resource utilizations:

1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:
d. Move the subsetting IF closer to the top of the DATA step. Make sure that you move the
subsetting IF to the appropriate location in the program.
e. Submit the revised program and record the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:
f. Were any resource(s) conserved when you moved the IF statement?

Level 2

2. Investigating the WORK System Option

a. In either the SAS Help facility (available from the Help menu or the Help icon in your
SAS session) or SAS OnlineDoc (available via a Web browser at
support.sas.com/documentation/index.html), find information about the WORK SAS
system option. The menu tree for finding the information in the SAS Help facility is shown
below:
Using SAS Software in Your Operating Environment Ö
<your operating system here> Ö Features of the SAS Language Ö
System Options under <your operating system here> Ö WORK
The menu tree for finding the information in the SAS OnlineDoc is shown below:
Base SAS Ö SAS 9.2 Companion for <your operating system here> HTML Ö
Features of the SAS Language for <your operating system here> (menu tree on left side) Ö
System Options under <your operating system here> Ö WORK System Option
b. Answer the following questions:
1) Can you submit the WORK system option in an OPTIONS statement? (YES or NO)
2) Explain why or why not.
 1.3 SAS DATA Step Processing 1-35

Level 3

3. Investigating the UTILLOC SAS System Option

In either the SAS Help facility (available from the Help menu or the Help icon in your SAS session)
or SAS OnlineDoc (available via a Web browser at support.sas.com/documentation/index.html),
find information about the UTILLOC SAS system option.
Answer the following questions:
a. What is the advantage of using the UTILLOC SAS system option?

b. Can both the WORK and UTILLOC SAS system options be specified for the same SAS session?
(YES or NO)
c. Explain your answer to part b.
1-36 Chapter 1 Introduction

1.4 Chapter Review

Chapter Review
1. What are the six resources consumed
by SAS programs?

2. What is the correct way to benchmark SAS programs?

3. What is a SAS data set page size?

75
 1.5 Solutions 1-37

1.5 Solutions

Solutions to Exercises
1. Benchmarking
a. Turn on the appropriate options for reporting the resource statistics in the log.
options fullstimer;

data order_fact;
<additional SAS code>
run;
b. Submit the program.
c. Record the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:

 The answers are specific to your operating environment.

1-38 Chapter 1 Introduction

d. Move the subsetting IF closer to the top of the DATA step. Make sure that you move the
subsetting IF to the appropriate location in the program.
data order_fact;
infile 'order_fact.dat' pad;
input @37 Order_Date date9. @;
if year(Order_Date)=2006;
input @1 Customer_ID 12.
@13 Employee_ID 12.
@25 Street_ID 12.
@46 Delivery_Date date9.
@55 Order_ID 12.
@67 Order_Type 2.
@69 Product_ID 12.
@81 Quantity 4.
@90 Total_Retail_Price 13.
@105 CostPrice_Per_Unit 10.
@115 Discount 5.;
format Customer_ID Employee_ID Street_ID Order_ID Product_ID 12.
Order_Date Delivery_Date date9.
Order_Type 2. Quantity 4. Total_Retail_Price dollar13.2
CostPrice_Per_Unit dollar10.2 Discount Percent.;
run;

proc print data=order_fact;

title 'Year 2006 Orders';
run;
e. Submit the revised program and record the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:

 The answers are specific to your operating environment.

f. Were any resource(s) conserved when you moved the IF statement?

You should see less CPU utilization when you move the IF statement closer to the top of the
program because, for observations not involved in the subset, numeric values are not
converted.
As the subset becomes smaller relative to the number of records in the raw data file, more
CPU is conserved.
 1.5 Solutions 1-39

2. Investigating the WORK System Option

a. In either the SAS Help facility or SAS OnlineDoc, find information about the WORK SAS
system option.
b. Answer the following questions:
1) Can you submit the WORK system option in an OPTIONS statement? (YES or NO) NO
2) Explain why or why not.
All file locations for SAS utilities must be set at SAS invocation, rather than in an
OPTIONS statement.
3. Investigating the UTILLOC SAS System Option
In either the SAS Help facility or SAS OnlineDoc, find information about the UTILLOC SAS system
option.
Answer the following questions:
a. What is the advantage of using the UTILLOC SAS system option?
You can avoid storing the utility swap files in the Work library, which frees space for
temporary SAS files. You can send the utility swap files to an under-utilized file system.
b. Can both the WORK and UTILLOC SAS system options be specified for the same SAS session?
(YES or NO) YES
c. Explain your answer to part b.
The WORK SAS system option points to the storage location for temporary SAS files. The
UTILLOC SAS system option specifies where utility swap files from SAS steps are stored.
The WORK SAS files are available to your entire SAS session unless they are specifically
deleted. The UTILLOC swap files are erased when the SAS step successfully completes.
1-40 Chapter 1 Introduction

Solutions to Student Activities (Polls/Quizzes)

1.02 Quiz – Correct Answer

Determine the path to use to obtain information about the
SAS component objects.
Information relevant to this
course can be found by
following these paths in the
SAS Help facility:

Contents tab
Æ SAS Products
Æ Base SAS
Æ SAS 9.2 Language
Reference Dictionary
Æ Dictionary of
Component
Object Language
Elements
16

1.06 Multiple Choice Poll – Correct Answer

Which of the following SAS programs should be
benchmarked?
a. A report that shows all the customers in the United
Kingdom in March 2006
b. A report that calculates trends in sales at the end
of every day for every department
c. A report showing the projected total cost of a 5%
cost-of-living increase in employee salaries for a
Human Resources project conducted on January 1,
2007
d. A yearly report that calculates the average sales
of a line of apparel for the clothing manager

42
 1.5 Solutions 1-41

1.08 Quiz – Correct Answer

Use one of the following to determine the page size
of the orion.customer_dim SAS data set:
„ the CONTENTS procedure

„ the DATASETS procedure

„ the SAS Explorer window

What is the page size of the SAS data set

orion.customer_dim?
16,384 bytes in Windows
24,576 bytes in UNIX
18,432 bytes in z/OS

p301a02
58
1-42 Chapter 1 Introduction

Solutions to Chapter Review

Chapter Review Answers

1. What are the six resources consumed
by SAS programs?
„ programmer time

„ network bandwidth

„ CPU

„ Memory

„ I/O

„ disk storage space

76 continued...

Chapter Review Answers

2. What is the correct way to benchmark SAS programs?
a. Turn on the system options to report resource
usage.
b. Test each technique in a separate SAS session.
c. Test only one technique or change at a time.
d. Run the test under final conditions.
e. Run each program three to five times and
average the results.
f. Exclude outliers.
g. Turn off the resource usage reporting options.

77 continued...
 1.5 Solutions 1-43

Chapter Review Answers

3. What is a SAS data set page size?
The size of the SAS data set page is the unit of
data transfer between the system buffers and the
SAS buffers in memory. The default transfer is one
data set page at a time.
The page size determines the amount of memory
that is used when data is read and written. The
number of pages effects the I/O.

78
1-44 Chapter 1 Introduction
Chapter 2 Controlling I/O
Processing and Memory

2.1 Controlling I/O................................................................................................................. 2-3

Exercises .............................................................................................................................. 2-13

2.2 Controlling Data Set Size ............................................................................................. 2-16

Exercises .............................................................................................................................. 2-26

2.3 Compressing SAS Data Sets ....................................................................................... 2-29

Exercises .............................................................................................................................. 2-43

2.4 Controlling Memory (Self-Study)................................................................................. 2-45

2.5 Controlling the Page Size and the Number of Available Buffers (Self-Study) ........ 2-53

2.6 Chapter Review............................................................................................................. 2-59

2.7 Solutions ....................................................................................................................... 2-60

Solutions to Exercises .......................................................................................................... 2-60

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 2-67

Solutions to Chapter Review ................................................................................................ 2-71

2-2 Chapter 2 Controlling I/O Processing and Memory
 2.1 Controlling I/O 2-3

2.1 Controlling I/O

Objectives
„ Describe the importance of conserving I/O.
„ List techniques for reducing I/O.

I/O (Review)
SAS programs typically perform the following tasks:
„ reading data sets sequentially

„ performing analysis and data manipulation

„ writing data sets sequentially or writing reports

 I/O is one of the most important factors

for optimizing performance.

4
2-4 Chapter 2 Controlling I/O Processing and Memory

Where Is I/O Measured? (Review)

I/O
measured
here

Input
SAS Buffers
Data

* Caches memory

PDV
Output Buffers ID Gender Country Name
SAS
Data
I/O
measured
here
* Windows and UNIX Only
5

Using the Operating Environment Cache

in Windows and UNIX
Windows and UNIX use a file-caching mechanism
in the background.
„ A file cache is an area in memory that holds recently
accessed data.
„ By default, SAS reads and writes data through the
operating environment file cache, not by direct I/O.
„ File caching is beneficial if the same data is used
more than once.
„ File caching adds overhead to sequential I/O.

6
 2.1 Controlling I/O 2-5

Techniques for Reducing I/O Operations

1. Minimize the number of variables and observations.
2. Reduce the number of times that the data is
processed.
3. Create a SAS data file when you process the same
raw data file repeatedly.
4. Use the SASFILE statement to process a small
SAS data set repeatedly.
5. Minimize the size of the SAS data set.
6. Use appropriate BUFSIZE= and/or BUFNO= options
for random or sequential access.

7 continued...

Techniques for Reducing I/O Operations

7. Bypass system file caching in Windows and UNIX.
8. Create views in programs that require intermediate
temporary SAS data files.
9. Create indexes on variables used for WHERE
processing.

8
8
2-6 Chapter 2 Controlling I/O Processing and Memory

Technique 1: Process Only the Necessary

Variables and Observations
Simple techniques can conserve I/O. The amount of I/O
saved depends on the size of the subset being
processed.
„ Reduce the number of variables.

– DROP or KEEP statements

– DROP= or KEEP= data set options
„ Reduce the number of observations.

– WHERE statement
– WHERE= data set option
– OBS= and FIRSTOBS= data set options

Subsetting Data
Program 1: Subsetting in the Procedure
One way to create a subset is to use the WHERE
statement in a procedure.
data bonus;
set orion.staff;
YrEndBonus=Salary*0.05;
run;
proc means data=bonus mean sum;
where Job_Title contains 'Manager';
class Manager_ID;
var YrEndBonus;
run;

 The data set bonus contains 11 variables

and 424 observations. p302d01
10
 2.1 Controlling I/O 2-7

Subsetting Data
Program 2: Subsetting in the DATA Step
Because the DATA step is required to create the variable
YrEndBonus, it is more efficient to subset in the DATA
step.
data bonus(keep=Manager_ID YrEndBonus);
set orion.staff(keep=Job_Title Salary Manager_ID);
where Job_Title contains 'Manager';
YrEndBonus=Salary*0.05;
run; I/O savings result
proc means data=bonus mean sum; from reducing the
class Manager_ID; number of variables
var YrEndBonus; and observations
run; in the input and
output data sets.
 The data set bonus contains two variables and 41
observations. p302d01
11

 Because of the way that SAS reads data, the savings in the DATA step are when the data set
bonus is output. Fewer variables and observations are written, so more can go on a single data set
page.

Using the KEEP=/DROP= Options

I/O
measured
here KEEP=/DROP= data set option
Input on the input data set
SAS Buffers
Data

memory
PDV
Job_ Salary Manager YrEnd
D D
Output Title _ID Bonus
Buffers
I/O
Data measured
Set here

KEEP=/DROP= data set option

on the output data set
(KEEP/DROP statement in the DATA step)
12
2-8 Chapter 2 Controlling I/O Processing and Memory

Setup for the Poll

p302d01 Program 1
data bonus;
set orion.staff;
YrEndBonus=Salary*0.05;
run; I/O savings results
proc means data=bonus mean sum;
where Job_Title contains 'Manager'; from reducing the
class Manager_ID; number of variables
var YrEndBonus; and observations
run;
in the input and
p302d01 Program 2 output data sets.
data bonus(keep=Manager_ID YrEndBonus);
set orion.staff(keep=Job_Title Salary Manager_ID);
where Job_Title contains 'Manager';
YrEndBonus=Salary*0.05;
run;
proc means data=bonus mean sum;
class Manager_ID;
var YrEndBonus;
run;

14

2.01 Multiple Choice Poll

In addition to the I/O decrease when the DATA step
creates bonus, where does Program 2 have additional
decrease of I/O?
a. Fewer variables are read into the program data vector
from orion.staff in Program 2 because of the KEEP=
data set option.
b. The PROC MEANS in Program 2 loads a smaller
version of bonus.
c. There is no additional decrease in I/O; all of the
decrease in I/O occurs when the data set bonus
is created by the DATA step.

15
 2.1 Controlling I/O 2-9

Technique 2: Reducing the Number of Times

that Data Is Processed
The following techniques reduce the number of times that
data is processed:
„ Subset data within a procedure step if possible.

„ Create SAS data files. SAS can process SAS data

files more efficiently than raw data files.
„ Use engines efficiently.

„ Use indexes.

„ Access data through SAS views.

17

Subsetting Data within a Procedure Step

Program 1: Subset in the DATA Step
You can subset in the DATA step first, and then execute
the procedure on a smaller data set.
data big_salaries;
set orion.staff;
where Salary > 50000;
run;

proc means data=big_salaries mean sum;

class Manager_ID;
var Salary;
run;

p302d02
18
2-10 Chapter 2 Controlling I/O Processing and Memory

Subsetting Data within a Procedure Step

Program 2: Subset in the Procedure
A better way is to execute only one step and subset
in that step.
proc means data=orion.staff mean sum;
class Manager_ID;
var Salary;
where Salary > 50000; I/O savings results
run; from avoiding an
extra step for
subsetting.

p302d02
19

Technique 3: Creating a SAS Data File

Program 1: Write two programs. Each program should
create a temporary SAS data set to read in
the same raw data file and create a report.
data prices;
infile 'prices.dat' dlm='*';
input Product_ID : 12. Start_Date : date9. End_Date : date9.
Unit_Cost_Price:dollar7.2 Unit_Sales_Price:dollar7.2;
run;
proc print data=prices(obs=5);
title1 "Prices Data Set";
run;

data prices;
infile 'prices.dat' dlm='*';
input Product_ID : 12. Start_Date : date9. End_Date : date9.
Unit_Cost_Price:dollar7.2 Unit_Sales_Price:dollar7.2;
run;
proc means data=prices(keep=Unit_Cost_Price Unit_Sales_Price);
var Unit_Cost_Price Unit_Sales_Price;
run;
p302d03
20
 2.1 Controlling I/O 2-11

Creating a SAS Data File

Program 2: Create a permanent SAS data set by reading
in the raw data file one time. Write two
programs using the SAS data set to create a
report.
data orion.prices;
infile 'prices.dat' dlm='*';
input Product_ID : 12. Start_Date : date9.
End_Date : date9.
Unit_Cost_Price : dollar7.2
Unit_Sales_Price : dollar7.2;
run; I/O savings from
proc print data=orion.prices(obs=5); reading the raw
title1 "Prices Data Set";
run; data once

proc means data=orion.prices

(keep=Unit_Cost_Price Unit_Sales_Price);
var Unit_Cost_Price Unit_Sales_Price;
run;
21 p302d03

Technique 4: SASFILE Global Statement

If your program uses the same data set multiple times,
the SASFILE statement can reduce I/O resources.
„ The SASFILE statement loads the SAS data set into
memory in its entirety, instead of a few pages at
a time.
„ After it is loaded, the data set is held in memory for
subsequent DATA and PROC step processing.
„ A second SASFILE statement closes the file and frees
the SAS buffers.
„ It is useful for data sets that fit entirely into memory.

„ The SASFILE statement should not be used if the data

set is larger than available physical memory.
The reduction in I/O resources comes
at the cost of increased memory usage.
22

 The SASFILE statement can also reduce execution (CPU) time.

2-12 Chapter 2 Controlling I/O Processing and Memory

Business Scenario
Create reports using the PRINT, TABULATE, MEANS,
and FREQUENCY procedures against a single SAS data
set.
sasfile orion.customer_dim load;

proc freq data=orion.customer_dim;

The orion.customer_dim data
tables Customer_Country Customer_Type;
run;
set is read into memory only
proc print data=orion.customer_dim noobs;
oncemembers
where Customer_Type='Orion Club Gold instead high
of four times. This
activity';
var Customer_ID Customer_Name Customer_Age_Group;
run; results in one-fourth as many
I/O operations,
proc means data=orion.customer_dim mean median max which
min; can also
var Customer_Age;
class Customer_Group;
reduce elapsed time. However,
run; it comes at the expense of
proc tabulate data=orion.customer_dim format=8.;
increased memory usage.
class Customer_Age_Group Customer_Type;
table Customer_Type All=Total,
Customer_Age_Group*n=' ' All=Total*n=' '/rts=45;
run;

sasfile orion.customer_dim close;

23 p302d04

SASFILE Global Statement

General form of the SASFILE statement:

SASFILE <libref.>member-name
<(password-data-set-option(s))>
OPEN | LOAD | CLOSE;

„ When the SASFILE statement executes, SAS

allocates the number of buffers based on the number
of pages in the SAS data set and index file.
„ If the file in memory increases in size during
processing by editing or appending data, the number
of buffers also increases.

24

OPEN opens the file and allocates the buffers, but defers reading the data into
memory until a procedure or a statement that references the file is executed.

LOAD opens the file, allocates the buffers, and reads the data into memory.

CLOSE frees the buffers and closes the file.

 2.1 Controlling I/O 2-13

Exercises

Level 1

1. Using the SASFILE Statement

a. Open the program p302e01 and submit it.
p302e01
proc print data=orion.organization_dim noobs;
where Department='Administration';
var Employee_ID Employee_Country Section Job_Title;
run;

proc means data=orion.organization_dim min mean max;

var Salary;
class Department;
run;

proc report data=orion.organization_dim headline headskip nowd;

column Company Department Employee_Hire_Date
Employee_BirthDate HiredAge;
define Company/order;
define Department/order;
define HiredAge/computed format=12.2 'Age when Hired';
compute HiredAge;
HiredAge=yrdif(Employee_BirthDate.sum,
Employee_Hire_Date.sum,'Act/Act');
endcomp;
run;

proc freq data=orion.organization_dim;

table Department*Company/norow nocol;
run;

proc means data=orion.organization_dim min mean max maxdec=2;

class Company;
var Salary;
run;
b. Note the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:
2-14 Chapter 2 Controlling I/O Processing and Memory

c. Add the appropriate statement(s) to open and load the entire orion.organization_dim data set into
memory. At the end of the program, close the data set.
d. Submit the revised program.
e. Note the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:
f. Which resources were conserved?

Level 2

2. Using Multiple SASFILE Statements

a. Open the program p302e02 and submit it.
p302e02
options fullstimer;
proc means data=orion.employee_donations sum mean median;
class Recipients;
var Qtr1-Qtr4;
run;

proc freq data=orion.employee_donations;

tables Recipients;
run;

proc report data=orion.employee_addresses headline headskip nowd;

columns Employee_ID Employee_Name City State;
define State/width=5;
where Country='US';
run;

proc freq data=orion.employee_addresses;

tables Country;
run;

proc sql;
select Employee_Name,
sum(Qtr1, Qtr2, Qtr3, Qtr4) as Total_Contribution,
Recipients
from orion.employee_addresses as a,
orion.employee_donations as d
where a.Employee_ID=d.Employee_ID;
quit;
options nofullstimer;
 2.1 Controlling I/O 2-15

b. Add the appropriate statement(s) to open and load both the orion.employee_addresses and
orion.employee_donations data sets into memory. At the end of the program, close the data sets.
c. Submit the revised program.

Level 3

3. Using the APPEND Procedure with the SASFILE Statement

a. Open the program p302e03 that contains a PROC COPY step to copy the data sets orion.sales
and orion.nonsales to your Work library so that you can maintain the integrity for those original
data sets. Submit the program.
p302e03
/*************************************/
/* the COPY procedure is creating a */
/* temporary copy of the data sets */
/* orion.sales and */
/* orion.nonsales */
/* so the integrity of the original */
/* data can be maintained for other */
/* demos and exercises. */
/*************************************/

proc copy in=orion out=work;

select sales nonsales;
run;

b. Add the appropriate statement(s) to open and load the entire work.sales data set into memory, a
PROC APPEND step to append the temporary work.nonsales data set to the temporary
work.sales data set, and a PROC PRINT step to print the work.sales data set. At the end of the
program, close the data set.
c. Submit the revised program.
2-16 Chapter 2 Controlling I/O Processing and Memory

2.2 Controlling Data Set Size

Objectives
„ List techniques to reduce data storage.
„ Describe how SAS stores numeric values.
„ Describe how to safely reduce the space required
to store numeric values in SAS data sets.

28

Techniques for Reducing Data Set Size

1. Store integers as reduced-length numerics.
2. Compress the data set.

 Reducing the size of a SAS data set reduces the I/O

required to process it.

29
 2.2 Controlling Data Set Size 2-17

Characteristics of Numeric Variables

Numeric variables have the following characteristics:
„ are stored as floating-point numbers in real-binary
representation
– store multiple digits per byte
– use a minimum of one byte to store the sign and
exponent of the value (depending on the operating
environment) and use the remaining bytes to store
the mantissa of the value
„ take 8 bytes of storage per variable, by default,
but can be reduced in size
„ always have a length of 8 bytes in the PDV

30

Default Length of Numeric Variables

The number 35,298 can be written as follows:

+0.35298*(10**5)
Sign Mantissa Base Exponent

SAS stores numeric variables in floating-point form:

Exponent Sign Mantissa

31

SAS stores numeric values in native floating-point representation.

Summary of Floating-Point Numbers Stored in 8 Bytes

Representation Base Exponent Bits Maximum Mantissa Bits

IBM mainframe 16 7 56

IEEE (UNIX, Windows, 2 11 52

Open VMS Alpha)
2-18 Chapter 2 Controlling I/O Processing and Memory

Assigning the Length of Numeric Variables

You can use a LENGTH statement to assign a length
from 2 to 8 bytes to numeric variables.
data emps_short;
length Street_ID 6
Employee_ID Manager_ID 5
Street_Number Employee_Hire_Date
Employee_Term_Date Birth_Date
Salary 4
Dependents 3;
merge employee_addresses
employee_organization
employee_payroll
employee_phones;
by Employee_ID;
run;

If the variable is numeric, the length applies only to the

output data set. If the variable is character, the length
applies to the program data vector and the output data
32 set. p302d05

Log
445 data emps_short;
446 length Street_ID 6
447 Employee_ID Manager_ID 5
448 Street_Number Employee_Hire_Date
449 Employee_Term_Date Birth_Date
450 Salary 4
451 Dependents 3;
452 merge employee_addresses
453 employee_organization
454 employee_payroll
455 employee_phones;
456 by Employee_ID;
457 run;

WARNING: Multiple lengths were specified for the BY variable Employee_ID by input data sets and
LENGTH, FORMAT, INFORMAT, or ATTRIB statements. This may cause unexpected results.
WARNING: Multiple lengths were specified for the variable Street_ID by input data set(s). This
may cause truncation of data.
WARNING: Multiple lengths were specified for the variable Street_Number by input data set(s).
This may cause truncation of data.
WARNING: Multiple lengths were specified for the variable Manager_ID by input data set(s). This
may cause truncation of data.
WARNING: Multiple lengths were specified for the variable Salary by input data set(s). This may
cause truncation of data.
WARNING: Multiple lengths were specified for the variable Birth_Date by input data set(s). This
may cause truncation of data.
WARNING: Multiple lengths were specified for the variable Employee_Hire_Date by input data
set(s). This may cause truncation of data.
WARNING: Multiple lengths were specified for the variable Employee_Term_Date by input data
set(s). This may cause truncation of data.
WARNING: Multiple lengths were specified for the variable Dependents by input data set(s). This
may cause truncation of data.

(Continued on the next page.)

 2.2 Controlling Data Set Size 2-19

NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_ADDRESSES.
NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_ORGANIZATION.
NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_PAYROLL.
NOTE: There were 923 observations read from the data set WORK.EMPLOYEE_PHONES.
NOTE: The data set WORK.EMPS_SHORT has 923 observations and 21 variables.
NOTE: DATA statement used (Total process time):
real time 0.18 seconds
cpu time 0.03 seconds

To decrease the length of all newly created numeric variables, you can use the DEFAULT= option in the
LENGTH statement:
data emps_short;
length default=4;
<additional SAS code>
run;

 The length of a character variable is determined by the first reference that creates the variable
when the DATA step is compiled. In addition to the LENGTH statements, character variables can
be created by using assignment statements, format statements, and read statements, for example
SET, MERGE, and INPUT statements.
2-20 Chapter 2 Controlling I/O Processing and Memory

Comparing Results
To determine whether the data sets emps_short
and emps are equivalent, you can use the COMPARE
procedure.

proc compare data=emps compare=emps_short;

run;

p302d06
33

General form of PROC COMPARE:

PROC COMPARE <option(s)>;

BY <DESCENDING> variable-1 ...
<DESCENDING> variable-n>
<NOTSORTED>;
ID <DESCENDING> variable-1
<...<DESCENDING> variable-n>
<NOTSORTED>;
VAR variable(s);
WITH variable(s);
RUN;

Task Statement

Compare the contents of SAS data sets, or compare PROC COMPARE

two variables.

Produce a separate comparison for each BY group. BY

Identify variables to use to match observations. ID

Restrict the comparison to values of specific VAR

variables.

Compare variables of different names. WITH and VAR

Compare two variables in the same data set. WITH and VAR
 2.2 Controlling Data Set Size 2-21

Comparing Data Sets

Partial PROC COMPARE Output
The COMPARE Procedure
Comparison of WORK.EMPS with WORK.EMPS_SHORT
(Method=EXACT)

Observation Summary

Observation Base Compare

First Obs 1 1
Last Obs 923 923

Number of Observations in Common: 923.

Total Number of Observations Read from WORK.EMPS: 923.
Total Number of Observations Read from WORK.EMPS_SHORT: 923.

Number of Observations with Some Compared Variables Unequal: 0.

Number of Observations with All Compared Variables Equal: 923.

NOTE: No unequal values were found. All values compared are exactly equal.

p302d06
34

 The COMPARE procedure is part of Base SAS.

Possible Storage Lengths for Integer Values

Windows and UNIX
Length Largest Integer
(bytes) Represented Exactly
3 8,192
4 2,097,152
5 536,870,912
6 137,438,953,472
7 35,184,372,088,832
8 9,007,199,254,740,992

35

The numbers are consecutive. For example, you can store numbers from -8192 to 8192 consecutively in
3 bytes on ASCII systems.
2-22 Chapter 2 Controlling I/O Processing and Memory

Possible Storage Lengths for Integer Values

z/OS
Length Largest Integer
(bytes) Represented Exactly
2 256
3 65,536
4 16,777,216
5 4,294,967,296
6 1,099,511,627,776
7 281,474,946,710,656
8 72,057,594,037,927,936

36

The numbers are consecutive. For example, you can store numbers from −256 continuously to 256 in
2 bytes on EBCDIC systems.

Assigning the Length of Numeric Variables

The use of a numeric length less than 8 bytes does the
following:
„ causes the number to be truncated to the specified
length when the value is written to the SAS data set
This reduces the number of bytes available for
the mantissa, which reduces the precision of
the number that can be accurately stored.
„ causes the number to be expanded to 8 bytes
in the PDV when the data set is read by padding
the mantissa with binary zeros

 Numbers are always 8 bytes in length in the PDV.

37

 SAS procedures also expand numbers to 8 bytes in memory.

 2.2 Controlling Data Set Size 2-23

Dangers of Reduced-Length Numeric Variables

It is not recommended that you change the length
of non-integer numeric variables.
data test;
length x 4;
X=1/10;
Y=1/10;
run;
data _null_;
set test;
put X=;
put Y=;
run;

p302a01
38

 In the same way that a decimal number system cannot store the fraction 1/3 exactly in a finite
number of digits, a binary number system (or multiple thereof, such as octal or hexadecimal)
cannot store the fraction 1/10 exactly in any finite number of digits.

2.02 Poll
Open the program p302a01 and submit it.
Look at the log.
Are the values of X and Y equal?
€ Yes
€ No

40
2-24 Chapter 2 Controlling I/O Processing and Memory

Numeric Precision
Partial SAS Log (Windows)
7 data test;
8 length x 4;
9 X=1/10;
10 Y=1/10;
11 run;

NOTE: The data set WORK.TEST has 1 observations and 2 variables.

NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

12
13 data _null_;
14 set test;
15 put X=;
16 put Y=;
17 run;

x=0.0999999642
y=0.1
NOTE: There were 1 observations read from the data set WORK.TEST.
NOTE: DATA statement used (Total process time):
real time 0.03 seconds
cpu time 0.00 seconds

42

Dangers of Reduced-Length Numeric Variables

It is not recommended that you reduce the length of
integer numeric variables inappropriately or that you
reduce the length of variables that hold large integer
numeric values. This example illustrates the effect of
inappropriately reducing integer values.
data test;
length X 3;
X=8193;
run;
data _null_;
set test;
put X=;
run;
p302d07
43

This example illustrates the dangers of inappropriately reducing integer values.

 2.2 Controlling Data Set Size 2-25

Numeric Precision
Partial SAS Log (Windows)
120 data test;
121 length X 3;
122 X=8193;
123 run;

NOTE: The data set WORK.TEST has 1 observations and 1

variables.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

124
125 data _null_;
126 set test;
127 put X=;
128 run;

x=8192
NOTE: There were 1 observations read from the data set
WORK.TEST.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

44

Advantages and Disadvantages

of Reduced-Length Numeric Variables
Advantages Disadvantages
Conserves data storage Uses additional CPU to
space read
Requires less I/O to read Can alter high-precision
values such as non-
integer and large integer
values

45
2-26 Chapter 2 Controlling I/O Processing and Memory

Exercises

Level 1

4. Creating Reduced Length Numerics

The data set orion.internet contains the orders placed over the Internet. The data set orion.retail
contains the items purchased in a store. The data set orion.catalog contains the items purchased from
the catalog.
a. Use PROC CONTENTS to determine the names and types of the variables in each of those data
sets.
b. Open the program p302e04. Edit the DATA step to concatenate the three data sets to create a data
set named all_customers. Appropriately reduce the length of the numeric variables using an
adequate length for the values of the variables.
p302e04
data all_customers;
set orion.catalog orion.internet orion.retail;
run;
c. Use PROC CONTENTS to check the length of the numeric variables.

Level 2

5. Creating Reduced Length Numerics and Precision

Open the program p302e05 and submit it. At what point in the sequence of numbers does the length
of the variable Num5 lose precision?
p302e05
data five;
length Num5 5 Num8 8;
do Num8=1e10 to 1e13 by 1e11;
Num5=Num8;
output;
end;
run;

proc print data=five;

title 'Reducing the length of numeric data to 5';
format Num5 Num8 20.;
run;
 2.2 Controlling Data Set Size 2-27

Level 3

6. Determining the Minimum Number of Bytes for Reduced Length Numerics

The following program from the Help facility finds the minimum length of bytes (MINLEN) needed
for numbers stored in a native SAS data set named numbers. The numbers data set contains the
variable Value. The variable Value contains a range of numbers. In this example, the range is 8191 to
8194 for Windows and UNIX and from 269 to 272 for z/OS.
p302e06
/* Windows and UNIX DATA step */
data numbers;
input Value;
datalines;
8191
8192
8193
8194
;

/* z/OS DATA step */

/*
data numbers;
input Value;
datalines;
269
270
271
272
;
*/
data temp;
set numbers;
X=Value;
do L=8 to 1 by -1;
if X NE trunc(X,L) then
do;
MinLen=L+1;
output;
return;
end;
end;
run;

title;
proc print noobs;
var Value MinLen;
run;
2-28 Chapter 2 Controlling I/O Processing and Memory

a. Run the program that is stored in p302e06 and examine the output.
b. Investigate the Help facility to determine why the minimum length for the number 8194 is less
than that of the number 8193 (Windows and UNIX) or why 272 is less than that of the number
271 (z/OS). The information can be found by following this path:
SAS Products Ö Base SAS Ö SAS Language References: Concepts Ö SAS System
Concepts Ö SAS Variables Ö Numeric Precision in SAS Software
 2.3 Compressing SAS Data Sets 2-29

2.3 Compressing SAS Data Sets

Objectives
„ Define the structure of a compressed SAS data file.
„ Create a compressed SAS data file.
„ List the advantages and disadvantages
of compression.

49

2.03 Poll
By default, the observations in a SAS data file have
varying lengths.
€ Yes
€ No

51
2-30 Chapter 2 Controlling I/O Processing and Memory

Simplified Uncompressed Data File Structure

24 / Obs Obs Obs Obs Obs 1
40 bit /
Page 1 2 3 4 5 * Descriptor
1 byte obs
OH OH

24 / Obs Obs Obs Obs Obs Obs Obs 1

Page 40 bit /
6 7 8 9 10 11 12
2 byte * obs
OH OH

. . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . .

24 / Obs Obs Obs 1

40 bit /
Page x y z * Unused space
byte obs
n
OH OH

53

 This is a visualization tool to help you understand how SAS data files are structured. SAS data
files are not actually stored in exactly this manner.

Uncompressed SAS Data File

The following are features of uncompressed
SAS data files:
„ All observations use the same number of bytes.

„ Each variable occupies the same number of bytes

in every observation.
„ Character values are padded with blanks.

„ Numeric values are padded with binary zeros.

„ The descriptor portion of the data set uses part

of the first data set page.

54 continued...
 2.3 Compressing SAS Data Sets 2-31

Uncompressed SAS Data File

„ There is a 24-byte overhead at the beginning of each
page on 32-bit systems.
„ There is a 40-byte overhead at the beginning of each
page on 64-bit systems.
„ There is a 1-bit per observation overhead, rounded
up to the nearest byte.
„ New observations are added at the end of the file.
If a new page is needed for a new observation,
a whole data set page is added.
„ Deleted observation space is never reused, unless
the entire data file is rebuilt.

55

In an uncompressed SAS data file, each observation is a fixed-length record.

Simplified Structure of a Compressed Data Set

24 | 12 | 24 O O
Obs Obs
Page 40 bytes/ b b
* Obs 7 Obs 6 Obs 5 3 2 Descriptor
1 byte obs s s
OH OH 4 1

24 | 12 | 24 O O Obs O
Page 40 bytes/ Obs b Obs b Obs b
* Obs 14 Obs 13 10
2 byte obs 16 s 12 s 9 s
OH OH 15 11 8

.
.
.

24 | 12 | 24 O
40 bytes/ b
Page byte obs * s Obs z
n OH OH y

* Unused space
56

This is a visual depiction of the storage used for a compressed SAS data file.
2-32 Chapter 2 Controlling I/O Processing and Memory

Compressed SAS Data File

Features of compressed SAS data files:
„ Each observation is a single string of bytes. Variable
types and boundaries are ignored.
„ Each observation can have a different length.

„ Consecutive repeating characters and numbers are

collapsed into fewer bytes.
„ If an updated observation is larger than its original
size, it is stored on either the same data set page or
on a different page with a pointer to the original page.
„ The descriptor portion of the data set is stored at the
end of the first data set page.

57 continued...

SAS data files, but not views, can be stored in compressed form.

Compressed SAS Data File

„ There is a 24-byte overhead at the beginning of each
page on 32-bit systems.
„ There is a 40-byte overhead at the beginning of each
page on 64-bit systems.
„ There is a 12-byte-per-observation overhead on 32-bit
systems.
„ There is a 24-byte-per-observation overhead on 64-bit
systems.
„ Deleted observation space can be reused if the
REUSE=YES data set or system option was turned
on when the SAS data file was compressed.

58

Compressing a file reduces the number of bytes required to represent each observation. In a compressed
file, each observation is a variable-length record.
 2.3 Compressing SAS Data Sets 2-33

Compressing SAS Files

There are two different algorithms that can be used
to compress files:
„ the RLE (Run Length Encoding) compression
algorithm (COMPRESS=YES | CHAR)
„ the RDC (Ross Data Compression) algorithm
(COMPRESS=BINARY)

The optimal algorithm depends on the

characteristics of your data.

59
2-34 Chapter 2 Controlling I/O Processing and Memory

Creating a Compressed Data File

To create a compressed data file, use the COMPRESS=
output data set option or system option.
General forms of the COMPRESS= options:

SAS-data-set(COMPRESS=NO | YES | CHAR | BINARY)

OPTIONS COMPRESS=NO | YES | CHAR | BINARY;

60

COMPRESS= Value Action

NO does not compress the data file (default)

CHAR | YES uses the run-length encoding (RLE) compression algorithm, which
compresses repeating consecutive bytes, such as trailing blanks or repeated
zeros

BINARY uses Ross Data Compression (RDC), which combines run-length encoding
and sliding window compression

 The COMPRESS= data set option overrides the COMPRESS= system option.

The COMPRESS= options interact with two other system or data set options, POINTOBS= and
REUSE=. See “COMPRESS= Data Set Option” in the dictionary of SAS language elements in
SAS® Language Reference: Dictionary in the Base SAS documentation or use the online Help facility
for additional information about these interactions.
 2.3 Compressing SAS Data Sets 2-35

Comparing Compression Methods

COMPRESS=YES | CHAR
„ is effective with character data that contains repeated
characters (such as blanks).
COMPRESS=BINARY
„ takes significantly more CPU time to uncompress
than COMPRESS=YES | CHAR
„ is more efficient with observations greater than
a 1000 bytes in length
„ can be very effective with numeric data

„ can be effective with character data that contains

patterns, rather than simple repetitions.

61

2.04 Quiz
Open the program p302a02.
1. Change the data set name to empchar. Add the
COMPRESS=CHAR data set option to the DATA
step and submit the program.
By what percentage was the data set reduced or
increased?
2. Change the data set name to empbin. Add the
COMPRESS=BINARY data set option to the DATA
step and submit the program.
By what percentage was the data set reduced or
increased?

63
2-36 Chapter 2 Controlling I/O Processing and Memory

Using the COMPRESS=CHAR Option

Partial SAS Log (Windows)
44
45 data empchar(compress=char);
46 merge employee_addresses employee_organization
47 employee_payroll employee_phones;
48 by Employee_ID;
49 run;

NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_ADDRESSES.
NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_ORGANIZATION.
NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_PAYROLL.
NOTE: There were 923 observations read from the data set WORK.EMPLOYEE_PHONES.
NOTE: The data set WORK.EMPCHAR has 923 observations and 21 variables.
NOTE: Compressing data set WORK.EMPCHAR decreased size by 60.71 percent.
Compressed is 11 pages; un-compressed would require 28 pages.
NOTE: DATA statement used (Total process time):
real time 0.04 seconds
cpu time 0.01 seconds

65

Using the COMPRESS=BINARY Option

Partial SAS Log (Windows)
50 data empbin(compress=binary);
51 merge employee_addresses employee_organization
52 employee_payroll employee_phones;
53 by Employee_ID;
54 run;

NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_ADDRESSES.
NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_ORGANIZATION.
NOTE: There were 424 observations read from the data set WORK.EMPLOYEE_PAYROLL.
NOTE: There were 923 observations read from the data set WORK.EMPLOYEE_PHONES.
NOTE: The data set WORK.EMPBIN has 923 observations and 21 variables.
NOTE: Compressing data set WORK.EMPBIN decreased size by 57.14 percent.
Compressed is 12 pages; un-compressed would require 28 pages.
NOTE: DATA statement used (Total process time):
real time 0.03 seconds
cpu time 0.03 seconds

67
 2.3 Compressing SAS Data Sets 2-37

Summary of Compression Results

Data Set Option Algorithm Number Decreased
Used Used of Bytes Size

employees None None 458,752 0%

empchar YES | CHAR RLE 180,224 60.71%
empbin BINARY RDC 196,608 57.14%

 Your results might differ, depending on SAS option

settings and the operating system.

68

The above comparison is for SAS 9.2 running on a Windows platform.

How SAS Compresses Data

A SAS data file has these variables:
Name Type Length
LastName Character 20
FirstName Character 15

In uncompressed form, all observations use 35 bytes

for these two variables.
Input Buffer 1 2 3
1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 5
A D A M S B i l l ...
LastName FirstName

69
2-38 Chapter 2 Controlling I/O Processing and Memory

Using COMPRESS=CHAR | YES

In run-length encoding compressed form, the LastName
and FirstName values for this observation use only 13
bytes.
1
1 2 3 4 5 6 7 8 9 0 1 2 3
@ A D A M S # @ B I L L #
LastName FirstName

„ The @ is a symbol that indicates how many

uncompressed characters follow.
„ The # is a symbol that indicates the number of blanks
repeated at this point in the observation.

70

Using COMPRESS=BINARY
Ross Data Compression uses both run-length encoding
and sliding window compression.
A SAS data set has these variables:
Name Type Length
Answer1 Numeric 8
...
Answer200 Numeric 8

In uncompressed form, the SAS data file resembles this:

Answer1 Answer2 Answer3 Answer4 Answer5 Answer6 ... Answer200
1 2 1 2 1 2 ... 2
1 1 1 1 1 1 ... 1
2 2 2 2 2 2 ... 2
71

Each uncompressed observation occupies 1600 bytes of storage.

 2.3 Compressing SAS Data Sets 2-39

Using COMPRESS=BINARY
In Ross Data Compression form, the first observation
in the data file resembles this:

1 2 3 4 5 6 7 8 9
+ +
@ 1 # @ 2 # %
1 1

„ The @ is a symbol that indicates how many

uncompressed characters follow.
„ The # is a symbol that indicates the number of binary
zeros repeated at this point in the observation.
„ The % is a symbol that indicates how many times
these values are repeated.

72

+
indicates the sign and exponent.
1

Compression Dependencies
Some data sets do not compress well or at all.

Because there is higher overhead for each observation,

a data file can occupy more space in compressed form
than in uncompressed form if the file has the following
characteristics:
„ few repeated characters

„ small physical size

„ few missing values

„ short text strings

73
2-40 Chapter 2 Controlling I/O Processing and Memory

Compression Dependencies
SAS Log (Windows)
1 data orders(compress=yes);
2 set orion.orders;
3 run;

NOTE: There were 490 observations read from the data set ORION.ORDERS.
NOTE: The data set WORK.ORDERS has 490 observations and 6 variables.
NOTE: Compressing data set WORK.ORDERS decreased size by 0.00 percent.
Compressed is 7 pages; un-compressed would require 7 pages.
NOTE: DATA statement used (Total process time):
real time 1.04 seconds
cpu time 0.12 seconds

55 data orders(compress=binary);
56 set orion.orders;
57 run;

NOTE: There were 490 observations read from the data set ORION.ORDERS.
NOTE: The data set WORK.ORDERS has 490 observations and 6 variables.
NOTE: Compressing data set WORK.ORDERS increased size by 28.57 percent.
Compressed is 9 pages; un-compressed would require 7 pages.
NOTE: DATA statement used (Total process time):
real time 0.09 seconds
cpu time 0.09 seconds p302d08
74

Compression Dependencies
When you use the COMPRESS= data set option or the
COMPRESS= system option, SAS knows the following:
„ the size of the overhead introduced by compression

„ the maximum size of an observation

If the maximum size of the observation is less than the

12-byte (32-bit systems) or 24-byte (64-bit system)
overhead introduced by compression, SAS does the
following:
„ disables compression

„ creates an uncompressed data set

„ issues a note stating that the file was not compressed

75
 2.3 Compressing SAS Data Sets 2-41

Compression Dependencies
SAS Log (Windows)
18 data test(compress=yes);
19 x=1;
20 run;

NOTE: Compression was disabled for data set WORK.TEST because

compression overhead would increase the size of the data set.
NOTE: The data set WORK.TEST has 1 observations and 1 variables.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

p302d09
76

Compression Trade-Offs
Uncompressed Compressed
Usually requires more disk Usually requires less disk
storage storage
Requires less CPU time to Requires more CPU time to
prepare an observation for prepare an observation for
I/O I/O
Uses more I/O operations Uses fewer I/O operations

The savings in I/O

operations greatly
outweigh the increase
in CPU time.
77 continued...
2-42 Chapter 2 Controlling I/O Processing and Memory

Compression Trade-Offs
Uncompressed Compressed
An updated observation fits in An updated observation might
its original location. be moved from its original
location.
Deleted observation space is Deleted observation space
never reused. can be reused.
New observations are always When REUSE=YES, new
inserted at the end of the observations might not be
data file. inserted at the end of the data
file.

78
 2.3 Compressing SAS Data Sets 2-43

Exercises

Level 1

7. Effects of Reading Compressed SAS Data Files

a. Submit the program p302e07.
b. Compare the user CPU time for reading employees, emps_short, empchar, and empbin.
employees:
emps_short:
empchar:
empbin:

Level 2

8. Compressing SAS Data Files

The following is the list of variables in the data set orion.product_list:
Alphabetic List of Variables and Attributes

# Variable Type Len Format Label

1 Product_ID Num 8 12. Product ID

4 Product_Level Num 8 12. Product Level
2 Product_Name Char 45 Product Name
5 Product_Ref_ID Num 8 12. Product Reference ID
3 Supplier_ID Num 8 12. Supplier ID

The following is the list of variables in the data set orion.supplier:

Alphabetic List of Variables and Attributes

# Variable Type Len Format Label

6 Country Char 2 Country

3 Street_ID Num 8 12. Street ID
5 Sup_Street_Number Char 8 Supplier Street Number
4 Supplier_Address Char 45 Supplier Address
1 Supplier_ID Num 8 12. Supplier ID
2 Supplier_Name Char 30 Supplier Name

a. You need to merge the two data sets together by Supplier_ID and create a compressed SAS data
set named supplier_names.
Which method of compression do you think would be the most appropriate?
2-44 Chapter 2 Controlling I/O Processing and Memory

b. Merge orion.product_list and orion.supplier to create a data set supplier_names. Compress it

using the method that you predicted in part a.
c. Merge orion.product_list and orion.supplier to create a data set supplier_names. Compress it
using the alternative method.
d. Which method was better?
e. Why was that method better?

 Your results might vary from the suggested solutions depending on the operating platform
and method used to create the data on that platform.

Level 3

9. Compressing a Library
a. Write a LIBNAME statement to assign the libref orcomp to the path as listed below. Use the
LIBNAME statement option COMPRESS=YES to compress the data sets that will be written to
that data library.

Operating Environment Library Location

Windows C:\temp

UNIX ~/temp

z/OS .prg3.tempdata

b. Write a PROC COPY step to copy data sets from the orion library to the orcomp library. The
PROC COPY step should copy only those data sets that begin with the letter "c". In addition,
ensure that you do not compress any of the data sets created in exercises after this one.
Hint: If you do not see compression messages the first time that you submit your code, look in the
Help facility or SAS OnlineDoc at the options for the PROC COPY statement.
c. Did any of them get larger? Yes or No
d. Why or why not?

e. Write a PROC DATASETS step with a DELETE statement to delete the library orcomp.
 2.4 Controlling Memory (Self-Study) 2-45

2.4 Controlling Memory (Self-Study)

Objectives
„ Investigate techniques for controlling memory.
„ Use system options to specify the amount of available
memory.

82

Comparing Memory, I/O, and CPU Resources

The techniques that reduce CPU and I/O can increase
memory usage.
„ Efficient use of the I/O subsystem uses larger buffers
and/or multiple buffers.
„ These buffers share memory space with the other
memory demands of your SAS session.

Benchmark carefully to balance the need to

conserve memory with the need to reduce CPU
and I/O.

83

Memory is a bigger issue on shared SAS systems (Windows servers, UNIX, z/OS) than on stand-alone
SAS systems (Windows PCs). Most individual SAS users will not encounter memory problems unless
their SAS programs use procedures with many distinct categorical values, perform sorts of large SAS data
sets, or use large in-memory lookup tables. In the first two cases, the swapping of utility files to disk
when physical memory is fully used can also increase the use of CPU and I/O resources.
2-46 Chapter 2 Controlling I/O Processing and Memory

Reducing Memory Usage

„ Use small data set page sizes when you create data
sets that will be accessed in a sparse, random pattern.
„ Use a single read buffer when the data is accessed
randomly instead of sequentially.
„ Use BY-group processing instead of CLASS
statements in those procedures that support both,
especially where you have pre-sorted data or can
use an existing index.

84

Using the BY Statement

Instead of using a CLASS statement to group data, you
can use the BY statement to specify the variables whose
values define the subgroup combinations for an analysis
by a SAS procedure.

85
 2.4 Controlling Memory (Self-Study) 2-47

Using the BY Statement

What are the differences between using a BY statement
and using a CLASS statement in a procedure?
BY Statement CLASS Statement
The data set must be sorted or The data set does not need to
indexed on the BY variables. be sorted or indexed on the
CLASS variables.
BY-group processing holds The CLASS statement
only one BY group in memory accumulates aggregates for all
at a time. CLASS groups simultaneously
in memory.
A percentage for the entire A percentage for the entire
report cannot be calculated report can be calculated with
with procedures such as the procedures such as the
REPORT or TABULATE REPORT or TABULATE
procedures. procedures.
86

PROC MEANS with a BY Statement

proc means data=orion.order_fact mean median
maxdec=2;
format Order_Date year4.;
by Order_Date;
var Quantity -- CostPrice_Per_Unit;
run;

p302d10
87
2-48 Chapter 2 Controlling I/O Processing and Memory

PROC MEANS with a BY Statement

Partial PROC MEANS Output with a BY Statement
----------------------Date Order was placed by Customer=2003--------------------

The MEANS Procedure

Variable Label Mean Median

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
Quantity Quantity Ordered 1.82 2.00
Total_Retail_Price Total Retail Price for This Product 177.59 105.95
CostPrice_Per_Unit Cost Price Per Unit 42.72 31.80
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

---------------------- Date Order was placed by Customer=2004 ----------------------

Variable Label Mean Median

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
Quantity Quantity Ordered 1.69 1.00
Total_Retail_Price Total Retail Price for This Product 146.13 85.65
CostPrice_Per_Unit Cost Price Per Unit 37.10 25.68
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

88

PROC MEANS with a CLASS Statement

proc means data=orion.order_fact mean median
maxdec=2;
format Order_Date year4.;
class Order_Date;
var Quantity -- CostPrice_Per_Unit;
run;

p302d10
89
 2.4 Controlling Memory (Self-Study) 2-49

PROC MEANS with a CLASS Statement

Partial PROC MEANS Output with a CLASS Statement
The MEANS Procedure
Date
Order
was
placed
by N
Customer Obs Variable Label Mean Median
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
2003 128 Quantity Quantity Ordered 1.82 2.00
Total_Retail_Price Total Retail Price for This Product 177.59 105.95
CostPrice_Per_Unit Cost Price Per Unit 42.72 31.80
2004 108 Quantity Quantity Ordered 1.69 1.00
Total_Retail_Price Total Retail Price for This Product 146.13 85.65
CostPrice_Per_Unit Cost Price Per Unit 37.10 25.68
2005 90 Quantity Quantity Ordered 1.70 1.00
Total_Retail_Price Total Retail Price for This Product 187.20 77.00
CostPrice_Per_Unit Cost Price Per Unit 49.45 25.20
2006 143 Quantity Quantity Ordered 1.57 1.00
Total_Retail_Price Total Retail Price for This Product 149.70 92.80
CostPrice_Per_Unit Cost Price Per Unit 44.25 29.95
2007 148 Quantity Quantity Ordered 1.93 2.00
Total_Retail_Price Total Retail Price for This Product 157.49 80.95
CostPrice_Per_Unit Cost Price Per Unit 37.53 21.35
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

90
2-50 Chapter 2 Controlling I/O Processing and Memory

2.05 Quiz
Open and submit the program p302a03. Answer
the following questions:
1. What is the advantage of technique 1?

2. What is the advantage of technique 2?

92

p302a03
options fullstimer;

/* Set up for TABULATE demo */

data order_fact(index=(Order_Date));
set orion.order_fact;
run;

/* Technique 1 */

proc tabulate data=order_fact;

format Order_Date year4.;
class Order_Date;
var Quantity -- CostPrice_Per_Unit;
tables Order_Date*(Quantity Total_Retail_Price CostPrice_Per_Unit),
n mean median pctsum;
run;

/* Technique 2 */

proc tabulate data=order_fact;

format Order_Date year4.;
by Order_Date;
var Quantity -- CostPrice_Per_Unit;
tables Quantity Total_Retail_Price CostPrice_Per_Unit,
n mean median pctsum;
run;

options nofullstimer;
 2.4 Controlling Memory (Self-Study) 2-51

Using Memory System Options

The operating environment will use disk-based virtual
memory when physical memory is depleted. This
adversely affects performance.

Use these SAS system options to limit the amount

of memory used by SAS:
„ REALMEMSIZE=

„ MEMSIZE=

Set the REALMEMSIZE= and MEMSIZE= options

to values below the amount of available physical
memory to prevent page thrashing or memory
swapping.

95

Consult the SAS OnlineDoc to see specifics for each operating environment.

Using System Options to Control Memory

The total amount of memory that a SAS job can consume
is limited by the MEMSIZE= invocation system option.

-MEMSIZE=n | n K | n M | n G | hexX | MIN | MAX

96

Consult the SAS OnlineDoc for specifics for each operating environment.
2-52 Chapter 2 Controlling I/O Processing and Memory

Using the MEMSIZE= Option

The MEMSIZE= option places a limit on the total amount
of memory that SAS dynamically allocates at any time.
This memory is supported by a combination of real
memory and space for utility files on disk.
Increase this option's value in small increments until
you find your optimal value.

 SAS does not automatically reserve or allocate

the amount of memory that you specify in the
MEMSIZE= system option. SAS will use only as
much memory as it needs to complete a process.

97

Using the REALMEMSIZE= System Option

Some SAS procedures use the REALMEMSIZE=
invocation option to specify how much memory *
the procedure can allocate and use without inducing
excessive page swapping.

-REALMEMSIZE=n | nK | nM | nG | hexX | MIN | MAX

The REALMEMSIZE= option should never be set beyond

the amount of real memory.
„ In UNIX and z/OS, REALMEMSIZE= specifies real
(physical) memory. In Windows, REALMEMSIZE=
specifies virtual memory.

98
 2.5 Controlling the Page Size and the Number of Available Buffers (Self-Study) 2-53

2.5 Controlling the Page Size and the Number of Available

Buffers (Self-Study)

Objectives
„ Control the page size of a SAS data set.
„ Use system and data set options to control memory
usage.
„ Describe the effect of operating environment caching.

101

Data Set Page Size (Review)

The page size of a SAS data set is specified in bytes. The
data set page is the unit of data transfer between the SAS
storage device and main memory.
I/O
measured The size of this buffer is the page
Input here size of the input data set.
SAS Buffers
Data
memory

PDV
Output Buffers
ID Gender Country Name
I/O
SAS
measured
Data here
The size of this buffer is the page
size of the output data set.
102
2-54 Chapter 2 Controlling I/O Processing and Memory

Controlling Page Size and Memory Usage

You can use the BUFSIZE= system option or data set
option to control the page size of an output SAS data set.

BUFSIZE= n | nK | nM | nG | nT | hexX | MIN | MAX

You can use the BUFNO= system option or data set

option to control the number of SAS buffers open
simultaneously in memory.

BUFNO= n

103

BUFSIZE= can only be used on output SAS data sets. BUFSIZE= sets the page size of a SAS data file,
which is a permanent attribute of the data set.
Increasing the BUFSIZE= option is sometimes useful for SAS data sets that are read sequentially (top to
bottom). Using a small BUFSIZE= value and BUFNO=1 is useful for SAS data sets that are read using
random access.
BUFSIZE= Value Specifies
n | nK | nM | nG | nT specifies the page size in multiples of 1 (bytes); 1,024 (kilobytes); 1,048,576
(megabytes); 1,073,741,824 (gigabytes); or 1,099,511,627,776 (terabytes). For
example, a value of 8 specifies 8 bytes, and a value of 3M specifies 3,145,728
bytes.
The default is 0, which causes SAS to use the minimum optimal page size for
the operating environment.
hexX specifies the page size as a hexadecimal value. You must specify the value
beginning with a number (0-9), followed by an X. For example, the value 2dx sets
the page size to 45 bytes.
MIN sets the page size to the smallest possible number in your operating environment,
down to the smallest four-byte, signed integer, which is -231-1, or approximately
-2 billion bytes.
This setting might cause unexpected results and should be avoided. Use
BUFSIZE=0 in order to reset the buffer page size to the default value in
your operating environment.
MAX sets the page size to the maximum possible number in your operating
environment, up to the largest four-byte, signed integer, which is 231-1, or
approximately 2 billion bytes.

 The specific values for the BUFSIZE= option depend on your operating environment.
 2.5 Controlling the Page Size and the Number of Available Buffers (Self-Study) 2-55

Controlling Page Size and Memory Usage

The product of BUFNO= and BUFSIZE= determines how
much data can be transferred in a read operation.
BUFSIZE BUFNO Bytes transferred
in one I/O

16384 2 32,768

 Increasing either BUFSIZE= or BUFNO= increases

the amount of data that can be transferred in a
read operation.

In Windows and UNIX, using the default BUFSIZE

and BUFNO is recommended.

104

Controlling Memory Usage

current SAS session
Page 3
Page 2
Page 1

bufno=3 data

 The buffer number is not a permanent attribute of the

data set and is valid only for the current step or
SAS session.
105
2-56 Chapter 2 Controlling I/O Processing and Memory

Using the Operating Environment Cache

in Windows and UNIX (Review)
„ A file cache is an area of memory that holds recently
accessed data.
„ By default, SAS reads and writes data through the
operating environment file cache, not by direct I/O.
„ The maximum I/O throughput rate depends on how
fast the operating environment file cache can process
the data.

106

For information about structuring RAID arrays and SAN arrays optimally for SAS, see
the white paper: Best Practices for Configuring your IO Subsystem for SAS®9 Applications
(support.sas.com/rnd/papers/sgf07/sgf2007-iosubsystem.pdf).

Using the SGIO System Option in Windows

The SGIO invocation system option performs the
following functions:
„ activates the Scatter-Read/Gather-Write I/O feature,
which disables file caching
„ improves I/O performance for SAS I/O files that
require a single sequential pass

General form of the SGIO system option:

NOSGIO | SGIO

107

The default is NOSGIO.

 SGIO is available as a data set option in SAS 9.2.

 2.5 Controlling the Page Size and the Number of Available Buffers (Self-Study) 2-57

Using the SGIO System Option in Windows

When SGIO is active, SAS does the following:
„ uses the number of buffers that are specified by the
BUFNO= system option to transfer data between disk
and RAM
„ bypasses Windows File Caching when reading
or writing data
„ reads ahead of the number of pages specified
by the BUFNO= system option and places the data
in memory before it is needed
When the data is needed, it is already in memory and
there is, in effect, a direct memory access.

Try different values of the BUFNO= system option

to tune each SAS job or DATA step.
108

Scatter-read/gather-write is active only for SAS I/O opened in INPUT or OUTPUT mode. If any SAS I/O
files are opened in UPDATE or RANDOM mode, SGIO is inactive for that process. Compressed and
encrypted files can also be read ahead using scatter-read/gather-write. I/O performance usually improves
as the value for the BUFNO increases.

Using Direct File I/O in UNIX

The ENABLEDIRECTIO and USEDIRECTIO LIBNAME
statement options, when used together, perform the
following functions:
„ activate direct I/O access

„ bypass UNIX file caching

„ improve I/O performance for SAS I/O files that require

a single sequential pass
The LIBNAME statement option enables direct I/O to any
file in the library. The data set option allows direct I/O to
the data set for this SAS program step only.

109

Data sets in the library specified without the USEDIRECTIO data set option use UNIX caching.
2-58 Chapter 2 Controlling I/O Processing and Memory

Using Direct File I/O in UNIX

General form of the ENABLEDIRECTIO and
USEDIRECTIO= LIBNAME statement options:

LIBNAME libref 'directory' USEDIRECTIO=NO|YES

ENABLEDIRECTIO;

The USEDIRECTIO option can also be a data set option.

General form of the USEDIRECTIO SAS data set option:

SAS-data-set-name (USEDIRECTIO=NO|YES)

110

For more information about using direct file I/O in UNIX, refer to the following:
support.sas.com/documentation/cdl/en/hostunx/61879/HTML/default/chloptfmain.htm
 2.6 Chapter Review 2-59

2.6 Chapter Review

Chapter Review
1. What is the purpose of the SASFILE statement?

2. Name the two techniques for reducing data set size.

3. What are the two types of SAS compression?

112
2-60 Chapter 2 Controlling I/O Processing and Memory

2.7 Solutions

Solutions to Exercises
1. Using the SASFILE Statement
a. Open the program p302e01 and submit it.
b. Note the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:

 The answers are specific to your operating environment.

c. Add the appropriate statement(s) to open and load the entire data set orion.organization_dim into
memory. At the end of the program, close the data set.
 2.7 Solutions 2-61

d. Submit the revised program.

p302s01
options fullstimer;
sasfile orion.organization_dim load;

proc print data=orion.organization_dim noobs;

where Department='Administration';
var Employee_ID Employee_Country Section Job_Title;
run;

proc means data=orion.organization_dim min mean max;

var Salary;
class Department;
run;

proc report data=orion.organization_dim headline headskip nowd;

column Company Department Employee_Hire_Date
Employee_BirthDate HiredAge;
define Company/order;
define Department/order;
define HiredAge/computed format=12.2 'Age when Hired';
compute HiredAge;
HiredAge=yrdif(Employee_BirthDate.sum,
Employee_Hire_Date.sum,'Act/Act');
endcomp;
run;

proc freq data=orion.organization_dim;

table Department*Company/norow nocol;
run;

proc means data=orion.organization_dim min mean max maxdec=2;

class Company;
var Salary;
run;

sasfile orion.organization_dim close;

options nofullstimer;
e. Note the following resource utilizations:
1) User CPU Time:
2) I/O:
(not applicable on Windows)
3) User Memory:

 The answers are specific to your operating environment.

2-62 Chapter 2 Controlling I/O Processing and Memory

f. Which resources were conserved?

User CPU time should decrease. If I/O is reported, there should be a significant drop in I/O.
2. Using Multiple SASFILE Statements
a. Open the program p302e02 and submit it.
b. Add the appropriate statement(s) to open and load both the orion.employee_addresses and
orion.employee_donations data sets into memory. At the end of the program, close the data sets.
c. Submit the revised program.
p302s02
options fullstimer;
sasfile orion.employee_addresses load;
sasfile orion.employee_donations load;

proc means data=orion.employee_donations sum mean median;

class Recipients;
var Qtr1-Qtr4;
run;

proc freq data=orion.employee_donations;

tables Recipients;
run;

proc report data=orion.employee_addresses headline headskip nowd;

columns Employee_ID Employee_Name City State;
where Country='US';
run;

proc freq data=orion.employee_addresses;

tables Country;
run;

proc sql;
select Employee_Name,
sum(Qtr1, Qtr2, Qtr3, Qtr4) as Total_Contribution,
Recipients
from orion.employee_addresses as a,
orion.employee_donations as d
where a.Employee_ID=d.Employee_ID;
quit;

sasfile orion.employee_addresses close;

sasfile orion.employee_donations close;
options nofullstimer;
 2.7 Solutions 2-63

3. Using the APPEND Procedure with the SASFILE Statement

a. Open the program p302e03.
b. Add the appropriate statement(s) to open and load the entire work.sales data set into memory, a
PROC APPEND step to append the temporary work.nonsales data set to the temporary
work.sales data set, and a PROC PRINT step to print the work.sales data set. At the end of the
program, close the data set.
c. Submit the revised program.
p302s03
/*************************************/
/* the COPY procedure is creating a */
/* temporary copy of the data sets */
/* orion.sales and */
/* orion.nonsales */
/* so the integrity of the original */
/* data can be maintained for other */
/* demos and exercises. */
/*************************************/

proc copy in=orion out=work;

select Sales NonSales;
run;

/*************************************/
/* The DATASETS procedure changes */
/* the names of FIRST and LAST to be */
/* compatable with the variables in */
/* the sales data. */
/*************************************/

sasfile sales load;

proc datasets lib=work;

modify nonsales;
rename First=First_Name
Last=Last_Name;
quit;

proc append base=sales data=nonsales force;

run;

proc print data=sales;

run;

sasfile sales close;

(Continued on the next page.)
2-64 Chapter 2 Controlling I/O Processing and Memory

/*************************************/
/* Alternative Solution */
/*************************************/

sasfile sales load;

proc append base=sales

data=nonsales (rename=(First=First_Name
Last=Last_Name)
force;
run;

proc print data=sales;

run;

sasfile sales close;

4. Creating Reduced Length Numerics
a. Use PROC CONTENTS to determine the names and types of the variables in each of the data
sets.
p302s04
proc contents data=orion.catalog;
run;

proc contents data=orion.internet;

run;

proc contents data=orion.retail;

run;
b. Open the program p302e04. Edit the DATA step to concatenate the three data sets to create a data
set named all_customers. Appropriately reduce the length of the numeric variables using an
adequate length for the values of the variables.
p302s04
/**************************************/
/* Do not reduce the length of */
/* the variables Total_Retail_Price, */
/* CostPrice_Per_Unit, or Discount. */
/* They are not integer values. */
/**************************************/

data all_customers;
length Quantity 3
Customer_ID Order_Date Delivery_Date 4
Employee_ID 5
Street_ID Order_ID 6
Product_ID 7;
set orion.catalog orion.internet orion.retail;
run;
 2.7 Solutions 2-65

c. Use PROC CONTENTS to check the length of the numeric variables.

p302s04
proc contents data=all_customers;
run;
5. Creating Reduced Length Numerics and Precision
Open the program p302e05 and submit it. At what point in the sequence of numbers does the length
of the variable Num5 lose precision?
The variable Num5 loses precision at the number 610000000000 on Windows and UNIX and at
1010000000000 on z/OS.
6. Determining the Minimum Number of Bytes for Reduced Length Numerics
a. Run the program that is stored in p302e06 and examine the output.
b. Investigate the Help facility to determine why the minimum length for the number 8194 is less
than that of the number 8193 (Windows and UNIX) or why 272 is less than that of the number
271 (z/OS).
The minimum length required for the value 271 (or 8193) is greater than the minimum
required for the value 272 (or 8194). This fact illustrates that it is possible for the largest
number in a range of numbers to require fewer bytes of storage than a smaller number. If
precision is needed for all numbers in a range, you should use the minimum length for all
the numbers, not only the largest one.

7. Effects of Reading Compressed SAS Data Files

a. Submit the program p302e07.
b. Compare the user CPU time for reading employees, emps_short, empchar, and empbin.

 CPU times vary by platform and other factors not controllable by SAS.

8. Compressing SAS Data Files

a. Which method of compression do you think would be the most appropriate?
Because the data set supplier_name is heavily character data, CHAR is probably most
appropriate.
b. Merge orion.product_list and orion.supplier to create a data set supplier_names. Compress it
using the method you predicted in part a.
p302s08
proc sort data=orion.product_list out=product_list;
by Supplier_ID;
run;
data sales(compress=char);
merge orion.supplier product_list;
by Supplier_ID;
run;
2-66 Chapter 2 Controlling I/O Processing and Memory

c. Merge orion.product_list and orion.supplier to create a data set supplier_names. Compress it

using the alternative method.
p302s08
proc sort data=orion.product_list out=product_list;
by Supplier_ID;
run;

data sales(compress=binary);
merge orion.supplier product_list;
by Supplier_ID;
run;
d. Which method was better?
CHAR
e. Why was that method better?
Heavily character data
9. Compressing a Library
a. Write a LIBNAME statement to assign the libref orcomp to the path as listed below. Use the
LIBNAME statement option COMPRESS=YES to compress the data sets that will be written to
that data library.
b. Write a PROC COPY step to copy data sets from the orion library to the orcomp library. The
PROC COPY step should copy only those data sets that begin with the letter "c". In addition,
ensure that you do not compress any of the data sets created in exercises after this one.
c. Did any of them get larger? Yes
d. Why or why not? Some of the data sets increased in size. They were not large data sets.
e. Write a PROC DATASETS step.
p302s09
libname orcomp 'C:\temp' compress=yes; /* Windows */
* libname orcomp '~/temp' compress=yes; /* UNIX */
* libname orcomp '.workshop.tempdata' compress=yes; /* z/OS */

proc copy in=orion out=orcomp noclone;

select c: ;
run;
proc datasets lib=orcomp;
delete c: ;
quit;
libname orcomp clear;
 2.7 Solutions 2-67

Solutions to Student Activities (Polls/Quizzes)

2.01 Multiple Choice Poll – Correct Answer

In addition to the I/O decrease when the DATA step
creates bonus, where does Program 2 have additional
decrease of I/O?
a. Fewer variables are read into the program data vector
from orion.staff in Program 2 because of the KEEP=
data set option.
b. The PROC MEANS in Program 2 loads a smaller
version of bonus.
c. There is no additional decrease in I/O; all of the
decrease in I/O occurs when the data set bonus
is created by the DATA step.

16

2.02 Poll – Correct Answer

Open the program p302a01 and submit it.
Look at the log.
Are the values of X and Y equal?
€ Yes
€ No

41
2-68 Chapter 2 Controlling I/O Processing and Memory

2.03 Poll – Correct Answer

By default, the observations in a SAS data file have
varying lengths.
€ Yes
€ No

By default, SAS observations have a fixed length.

52

2.04 Quiz – Correct Answer

Open the program p302a02.
1. Change the data set name to empchar. Add the
COMPRESS=CHAR data set option to the DATA
step and submit the program.
By what percentage was the data set reduced or
increased?
data empchar(compress=char);
merge employee_addresses
employee_organization
employee_payroll
employee_phones;
by Employee_ID;
run;

64 continued...
 2.7 Solutions 2-69

2.04 Quiz – Correct Answer

Open the program p302a02.
2. Change the data set name to empbin. Add the
COMPRESS=BINARY data set option to the DATA
step and submit the program.
By what percentage was the data set reduced or
increased?
data empbin(compress=binary);
merge employee_addresses
employee_organization
employee_payroll
employee_phones;
by Employee_ID;
run;

66 continued...

2.05 Quiz – Correct Answer

Open and submit the program p302a03. Answer
the following questions:
1. What is the advantage of technique 1?
Technique 1 enables the correct use of the
PCTSUM statistic.

2. What is the advantage of technique 2?

93
2-70 Chapter 2 Controlling I/O Processing and Memory

2.05 Quiz – Correct Answer

Open and submit the program p302a03. Answer
the following questions:
1. What is the advantage of technique 1?
Technique 1 enables the correct use of the
PCTSUM statistic.

2. What is the advantage of technique 2?

Technique 2 uses much less memory, but the
results for the PCTSUM statistic are not what you
want.

94
 2.7 Solutions 2-71

Solutions to Chapter Review

Chapter Review – Correct Answers

1. What is the purpose of the SASFILE statement?
The SASFILE statement is used to load an entire
SAS data set into memory and hold it there for
subsequent DATA or PROC steps to process.
2. Name the two techniques for reducing data set size.
Storing integer numbers as reduced length
numerics and SAS data compression
3. What are the two types of SAS compression?
CHAR (YES) and BINARY

113
2-72 Chapter 2 Controlling I/O Processing and Memory
Chapter 3 Accessing Observations

3.1 Creating an Index ........................................................................................................... 3-3

Exercises .............................................................................................................................. 3-28

3.2 Using an Index .............................................................................................................. 3-29

Exercises .............................................................................................................................. 3-47

3.3 Creating a Sample Data Set (Self-Study) .................................................................... 3-50

Exercises .............................................................................................................................. 3-70

3.4 Chapter Review............................................................................................................. 3-71

3.5 Solutions ....................................................................................................................... 3-72

Solutions to Exercises .......................................................................................................... 3-72

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 3-80

Solutions to Chapter Review ................................................................................................ 3-86

3-2 Chapter 3 Accessing Observations
 3.1 Creating an Index 3-3

3.1 Creating an Index

Objectives
„ Define indexes.
„ List the uses of indexes.
„ Use the DATA step to create indexes.
„ Use PROC DATASETS to create and maintain
indexes.
„ Use PROC SQL to create and maintain indexes.

3.01 Multiple Answer Poll

Do any of the data files that you use have indexes?
a. Yes, my SAS data sets have indexes.
b. Yes, I use data from an RDBMS (such as Oracle,
Teradata, Sybase, or DB2) that has indexes.
c. No, none of the data that I use has indexes.

5
3-4 Chapter 3 Accessing Observations

Using Indexes
An index is an optional file that you can create for
a SAS data file that does the following:
„ points to observations based on the values of one
or more key index variables
„ provides direct access to specific observations

 An index locates an observation by value.

Simplified Index File

The index file consists of entries that are organized
in a tree structure and connected by pointers.
Partial Listing of Simplified Index
orion.sales_history Customer_ID Record Identifier (RID)
Key Value Page(obs, obs, ...)
Customer_ID Employee_ID . . .
14958 121031 . . . 4006 17(85)
14844 121042 . . . 4021 17(89)
14864 99999999 . . .
4059 17(90)
14909 120436 . . .
14862 120481 . . . 4063 17(80, 86)
14853 120454 . . . .
14838 121039 . . . .
14842 121051 . . . .
14815 99999999 . . . 14958 1(1, 24)
14797 120604 . . . 14972 1(14)
. . . .
. . . .
. . . .
7

 The index is stored with the key values in ascending sorted order.
 3.1 Creating an Index 3-5

The Purpose of Indexes

Indexes can provide direct access to observations
in SAS data sets to accomplish the following:
„ yield faster access to small subsets (WHERE)

„ return observations in sorted order (BY)

„ perform table lookup operations (SET with KEY=)

„ join observations (PROC SQL)

„ modify observations (MODIFY with KEY=)

Why Use an Index?

How is data processed if the input data is not indexed?

data customer14958;
set orion.sales_history;
where Customer_ID=14958;
run;

9
3-6 Chapter 3 Accessing Observations

Reading SAS Data Sets without an Index

Input
SAS Buffers The WHERE statement
selects observations
Data by reading data
Data sequentially.
pages are
loaded. PDV

ID Gender Country Name

Output Buffers
SAS
Data

15

Why Use an Index?

How is data processed if the input data is indexed?

data customer14958;
set orion.sales_history;
where Customer_ID=14958;
run;

16
 3.1 Creating an Index 3-7

Reading SAS Data Sets with an Index

Index Index The index file

is checked.

Input
SAS Buffers The WHERE statement
Data selects observations
Only by using direct access.
necessary
pages are PDV
loaded.
ID Gender Country Name
Output Buffers
SAS
Data
23

When SAS uses an index to process data, SAS does the following:
• performs a binary search on the index file
• positions the index to the first entry containing a qualified value
• transfers a page of data containing the first record identifier for the qualified value to a buffer
• directly accesses the value specified by the record identifier
• positions the index to the next entry containing a qualified value
• transfers the page of data, if it is not already in the buffer
• directly accesses the value specified by the record identifier
• continues to process the data until there is no more data that satisfies the WHERE expression

 If the stored data values are sorted in ascending order by the indexed variables, fewer I/O
operations are required. If the data is not sorted on the index key values, but observations with the
same key values are near each other in the file, I/O will be minimized.
3-8 Chapter 3 Accessing Observations

Number of Index Buffers (Self-Study)

The buffer size of a SAS index is the unit of data transfer
between the SAS storage device and main memory.

Index
Index
Buffer

Input
SAS Buffers
Data

PDV
ID Gender Country Name
Output Buffers
SAS
Data

24
 3.1 Creating an Index 3-9

Controlling the Number of Index Buffers

(Self-Study)
You can use the IBUFNO= system option to control the
number of index buffers that are simultaneously open
in memory.

IBUFNO=n | nK | nM | nG | nT

25

SAS automatically allocates a minimal number of buffers in order to navigate the index file. Typically,
you do not need to specify extra buffers. However, using IBUFNO= to specify extra buffers can improve
execution time by limiting the number of input/output operations that are required for a particular index
file. However, the improvement in execution time comes at the expense of increased memory
consumption.

 Whereas too few buffers allocated to the index file decrease performance, over-allocation of
index buffers creates performance problems as well. Experimentation is the best way to determine
the optimal number of index buffers. For example, experiment with IBUFNO=3, then
IBUFNO=4, and so on, until you find the least number of buffers that produces satisfactory
performance results.

IBUFSIZE= Value Specifies

n | nK | nM | nG | nT specifies the number of extra index buffers to be allocated in multiples of 1

(bytes); 1,024 (kilobytes); 1,048,576 (megabytes); 1,073,741,824 (gigabytes); or
1,099,511,627,776 (terabytes). For example, a value of 8 specifies eight buffers,
and a value of 3k specifies 3,072 buffers.

The maximum value is 10,000.

hexX specifies the number of extra index buffers as a hexadecimal value. You must
specify the value beginning with a number (0-9), followed by an X.

MIN sets the number of extra index buffers to 0. This is the default.

MAX sets the number of extra index buffers to 0. This is the default
3-10 Chapter 3 Accessing Observations

How Is the Index File Checked?

Index Index The index file

is checked.

Input When an index is used,

SAS Buffers a binary search is done
Data on the index file.

PDV

ID Gender Country Name

Output Buffers
SAS
Data
26

Using a Binary Search

Partial Listing of Simplified Index File
orion.sales_history
Customer_ Employee_ Customer_ Record Identifier (RID)
. . . ID
RID ID ID Page(obs, obs, ...)
1 14958 121031 . . . Key Value

2 14844 121042 . . . 4006 17(85)

3 14864 99999999 . . . 4021 17(89)
4 14909 120436 . . .
4059 17(90)
. . .
. . . . . . 4063 17(80, 86)
where Customer_ID=14958;
. . .
.
22 14918 120918 . . . .
23 14844 121042 . . . .

24 14958 121031 . . . 14958 1(1, 24)

25 14821 120918 . . . 14972 1(14)
. . .
.
. . . . . .
.
. . .
.

27 ...
 3.1 Creating an Index 3-11

Using a Binary Search

Partial Listing of Simplified Index File
orion.sales_history
Customer_ Employee_ Customer_ Record Identifier (RID)
. . . ID
RID ID ID Page(obs, obs, ...)
1 Is14958
14958121031
in the . . . Key Value

2 top half or
14844 the
121042 . . . 4006 17(85)
3
bottom99999999
14864 half? . . . 4021 17(89)
4 14909 120436 . . .
4059 17(90)
. . .
. . . . . . 4063 17(80, 86)
where Customer_ID=14958;
. . .
.
22 14918 120918 . . . .
23 14844 121042 . . . .

24 14958 121031 . . . 14958 1(1, 24)

25 14821 120918 . . . 14972 1(14)
. . .
.
. . . . . .
.
. . .
.

28 ...

The binary search essentially divides the index file in half and asks, “Is the key value that I am searching
for above or below the halfway point?” The binary search continues to divide the remaining portions of
the index file in half until the key value is found.

Using a Binary Search

Partial Listing of Simplified Index File
orion.sales_history
Customer_ Employee_ Customer_ Record Identifier (RID)
. . . ID
RID ID ID Page(obs, obs, ...)
1 Is14958
14958121031
in the . . . Key Value

2 top half or
14844 the
121042 . . . 4006 17(85)
3
bottom half?
14864 99999999 . . . 4021 17(89)
4 14909 120436 . . .
4059 17(90)
. . .
. . . . . . 4063 17(80, 86)
where Customer_ID=14958;
. . .
.
22 14918 120918 . . . .
23 14844 121042 . . . .

24 14958 121031 . . . 14958 1(1, 24)

25 14821 120918 . . . 14972 1(14)
. . .
.
. . . . . .
.
. . .
.

29 ...
3-12 Chapter 3 Accessing Observations

Using a Binary Search

Partial Listing of Simplified Index File
orion.sales_history
Customer_ Employee_ Customer_ Record Identifier (RID)
. . . ID
RID ID ID Page(obs, obs, ...)
1 14958 121031 . . . Key Value

2 14844 121042 . . . 4006 17(85)

3 14864 99999999 . . . 4021 17(89)
4 14909 120436 . . .
4059 17(90)
. . .
. . . . . . 4063 17(80, 86)
where Customer_ID=14958;
. . .
.
22 14918 120918 . . . .
23 14844 121042 . . . .

24 14958 121031 . . . 14958 1(1, 24)

25 14821 120918 . . . 14972 1(14)
. . .
.
. . . . . .
.
. . .
.

30 ...

Using a Binary Search

Partial Listing of Simplified Index File
orion.sales_history
Customer_ Employee_ Customer_ Record Identifier (RID)
. . . ID
RID ID ID Page(obs, obs, ...)
1 14958 121031 . . . Key Value

2 14844 121042 . . . 4006 17(85)

3 14864 99999999 . . . 4021 17(89)
4 14909 120436 . . .
4059 17(90)
. . .
. . . . . . 4063 17(80, 86)
where Customer_ID=14958;
. . .
.
22 14918 120918 . . . .
23 14844 121042 . . . .

24 14958 121031 . . . 14958 1(1, 24)

25 14821 120918 . . . 14972 1(14)
. . .
.
. . . . . .
.
. . .
.

31
 3.1 Creating an Index 3-13

3.02 Multiple Choice Poll

If a WHERE statement uses an index to retrieve a small
subset of data, which of these resources is conserved?
a. I/O
b. Disk space
c. Memory
d. Programmer time

33

Business Scenario
The SAS data set orion.sales_history is often queried
with a WHERE statement.

Partial Listing of orion.sales_history

Customer Order_ Order_ Product_
. . . Product_ID ... . . .
_ID ID Type Group

14958 . . . 1230016296 1 210200600078 . . . N.D. Gear, Kids . . .

Eclipse
14844 . . . 1230096476 1 220100100354 . . . . . .
Clothing
14864 . . . 1230028104 2 240600100115 . . . Bathing Suits . . .
14909 . . . 1230044374 1 240100200001 . . . Darts . . .
14862 . . . 1230021668 1 240500200056 . . . Running Clothes . . .
14853 . . . 1230021653 1 220200200085 . . . Shoes . . .
14838 . . . 1230140184 1 220100300042 . . . Knitwear . . .
14842 . . . 1230025285 1 240200100053 . . . Golf . . .
14815 . . . 1230109468 3 230100700004 . . . Tents . . .
14797 . . . 1230168587 1 220101000004 . . . Shorts . . .

35
3-14 Chapter 3 Accessing Observations

Business Scenario
You need to create three indexes on the most frequently
used subsetting columns.
Index Name Index Variables
Customer_ID Customer_ID
Product_Group Product_Group

SaleID Order_ID
Product_ID
Partial Listing of orion.sales_history
Customer Order_ Order_ Product_
. . . Product_ID . . . . . .
_ID ID Type Group

14958 . . . 1230016296 1 210200600078 . . . N.D. Gear, Kids . . .

Eclipse
14844 . . . 1230096476 1 220100100354 . . . . . .
Clothing

36

Creating an Index
Customer_ID
Customer_ID
Order_ID
Product_Group
Product_ID
SaleID
Product_Group

Key variables in Indexes in the index

orion.sales_history file for orion.sales_history
sales_history.sas7bdat sales_history.sas7bndx
Directory-based Index File Naming Conventions
Index Name Index Variables Index Type
Customer_ID Customer_ID Simple
Product_Group Product_Group Simple
SaleID Order_ID Composite
Product_ID
37
 3.1 Creating an Index 3-15

Index Terminology
There are two types of indexes.

Type Based On Name Example

Simple the value of only automatically Customer_ID
one variable given the same
name as its key
Product_Group
variable
Composite the values of more must be given a SaleID=
than one variable name that is not (Order_ID
concatenated to the same as
Product_ID)
form a single any variable or
value existing index

38

Index Terminology
Index options include the following:
UNIQUE Values of the key variable(s) must be unique. This
option prevents an observation with a duplicate value
for the key variable(s) from being added to the data set.
Partial Listing of orion.sales_history
Customer Employee_ Order_
. . . Order_ID Product_ID Quantity . . .
_ID ID Type
14958 121031 . . . 1230016296 1 210200600078 1 . . .
14844 121042 . . . 1230096476 1 220100100354 1 . . .
14864 99999999 . . . 1230028104 2 240600100115 1 . . .
14909 120436 . . . 1230044374 1 240100200001 1 . . .
14862 120481 . . . 1230021668 1 240500200056 1 . . .
14853 120454 . . . 1230021653 1 220200200085 3 . . .
14838 121039 . . . 1230140184 1 220100300042 4 . . .

The concatenation of the values for Order_ID and

39 Product_ID forms a unique identifier for a row of data.

In an existing data set, if the variable(s) on which you attempt to create a unique index has duplicate
values, the index is not created and an error message is written to the SAS log.
3-16 Chapter 3 Accessing Observations

Index Terminology
Index options include the following:
NOMISS excludes all observations with missing values from the
index. Observations with missing values can still be
read from the data set, but not using the index.

„ If there is a large number of missing values for the key

variable(s), the NOMISS option can create a smaller
index file.
„ An index created with the NOMISS option is not used
for the following processing:
– a BY statement
– a WHERE expression satisfied by missing values

NOMISS cannot be used when you create indexes

using PROC SQL.
40

3.03 Multiple Answer Poll

On which of these indexed variables can you assign the
UNIQUE option?
a. Customer_ID in an orders data set where
a customer can place multiple orders
b. Order_Date in an orders data set
c. Employee_ID in a data set containing each
individual employee and the family members’
names stored in variables Dependent1 –
Dependent10
d. Product_ID in a data set containing the
product identifier and the product description

42
 3.1 Creating an Index 3-17

Creating Indexes
To create indexes at the same time that you create
a data set, use the INDEX= data set option on the output
data set.
To create or delete indexes on existing data sets,
use one of the following:
„ DATASETS procedure

„ SQL procedure

44

Creating Indexes
When you create the index, do the following:
„ designate the key variable(s)

„ specify the UNIQUE and/or the NOMISS

index option if appropriate
„ select a valid SAS name for the index
(composite index only)

A data set can have these index features:

„ multiple simple and composite indexes

„ character and numeric key variables

45
3-18 Chapter 3 Accessing Observations

Creating an Index with the

INDEX= Data Set Option
options msglevel=i;
data orion.sales_history(index=
(Customer_ID Product_Group
SaleID=(Order_ID Product_ID)/unique));
set orion.history;
Value_Cost=CostPrice_Per_Unit*Quantity;
Year_Month=mdy(Month_Num, 15, input(Year_ID,4.));
format Value_Cost dollar12.
Year_Month monyy7.;
label Value_Cost="Value Cost"
Year_Month="Month/Year";
run;

The following code would delete the indexes:

data orion.sales_history;
set orion.sales_history;
run;

46 p303d01
 3.1 Creating an Index 3-19

Creating an Index with the

INDEX= Data Set Option
General form of the INDEX= data set option:

SAS-data-file-name (INDEX =
(index-specification-1</option> </option>
…<index-specification-n</option> </option> >));

 For increased efficiency, use the INDEX= option to

create indexes when you initially create a SAS data
set.

47

The following are conditions for an index-specification:

simple index the name of the key variable

composite index index-name=(list of key variables)

You can specify the UNIQUE and/or the NOMISS option with the INDEX= data set option. Each option
is preceded by a slash (/).
The INDEX= data set option can also be used in procedures with OUT= options and with ODS OUTPUT
statements.
3-20 Chapter 3 Accessing Observations

Viewing Information about Indexes

To display information in the log concerning index
creation or index usage, change the value of the
MSGLEVEL= system option from its default value
of N to I.
General form of the MSGLEVEL= system option:
OPTIONS MSGLEVEL=N | I;

11 options msglevel=i;
12 data orion.sales_history(index=
13 (Customer_ID Product_Group
14 SaleID=(Order_ID
15 Product_ID)/unique));
16 set orion.sales_history;
17 run;

NOTE: There were 1500 observations read from the data set ORION.SALES_HISTORY.
NOTE: The data set ORION.SALES_HISTORY has 1500 observations and 22 variables.
NOTE: Composite index SaleID has been defined.
NOTE: Simple index Product_Group has been defined.
NOTE: Simple index Customer_ID has been defined.
48

N prints notes, warnings, and error messages. This is the default.

I prints informational or INFO notes that pertain to index creation and usage,
merge processing, host sort utilities, and threading in addition to notes,
warnings, and error messages.

Creating an Index with the

INDEX= Data Set Option
Advantages Disadvantages

You can create the data set To create an additional

and the index in one step. index, you must re-create
the existing indexes.
SAS only reads the data You need to know in
once. advance that indexes
are needed.

49
 3.1 Creating an Index 3-21

Managing Indexes with PROC DATASETS

options msglevel=n;
proc datasets library=orion nolist;
modify sales_history;
index create Customer_ID;
index create Product_Group;
index create SaleID=(Order_ID
Product_ID)/unique;
quit;
The following code would delete the indexes:
proc datasets library=orion nolist;
modify sales_history;
index delete Customer_ID
Product_Group SaleID;
quit;
p303d02
50

 The value for the MSGLEVEL= SAS system option is set to n because PROC DATASETS
issues its own notes.
3-22 Chapter 3 Accessing Observations

Managing Indexes with PROC DATASETS

You can use the DATASETS procedure on existing data
sets to create or delete indexes.
General form of the PROC DATASETS step to delete
or create indexes:

PROC DATASETS LIBRARY=libref NOLIST;

MODIFY SAS-data-set-name;
INDEX DELETE index-name;
INDEX CREATE index-specification
< / options>;
QUIT;

51

The INDEX CREATE statement in PROC DATASETS cannot be used if the index to be created already
exists.
If the index to be created already exists, you must do the following:
• delete the existing index of the same name
• create the new index
If you delete and create indexes in the same step, delete indexes first so that the newly created indexes can
reuse the space of the deleted indexes.
You can specify the UNIQUE or NOMISS option in the INDEX CREATE statement.
 3.1 Creating an Index 3-23

3.04 Quiz
Open and submit the program p303a01.
What error messages are in the log?

p303a01
options msglevel=n;
proc datasets library=orion nolist;
modify sales_history;
index create Customer_ID;
index create Product_Group;
index create SaleID=(Order_ID
Product_ID)/unique;
quit;

53

Managing Indexes with PROC DATASETS

Advantages Disadvantages

Additional indexes can be You can only create

created without re-creating indexes on existing SAS
the original indexes. data sets and existing
variables.
One or more indexes can PROC DATASETS cannot
be deleted without deleting perform data manipulation.
all of the indexes on the
data set.
If an index exists, it must
be deleted before it can be
re-created.
55
3-24 Chapter 3 Accessing Observations

Managing Indexes with PROC SQL

options msglevel=n; Name of Index
proc sql;
create index Customer_ID
on orion.sales_history(Customer_ID); Variable
create index Product_Group Name
on orion.sales_history(Product_Group);
create unique index SaleID
on orion.sales_history(Order_ID, Product_ID);
quit;

The following code would delete the indexes:

proc sql;
drop index Customer_ID, Product_Group, SaleID
from orion.sales_history;
quit;

p303d03
56

 The value for the MSGLEVEL= SAS system option is set to n because PROC SQL issues its own
notes.
 3.1 Creating an Index 3-25

Managing Indexes with PROC SQL

You can use PROC SQL on existing data sets to create
or delete indexes.
General form of the PROC SQL step to create or delete
indexes:

PROC SQL;
DROP INDEX index-name
FROM table-name;
CREATE <option> INDEX index-name
ON table-name(column-name-1,...
column-name-n);
QUIT;

57

PROC SQL cannot be used if the index to be created already exists.

If the index to be created already exists, you must do the following:
• drop the existing index of the same name
• create the new index
The SQL procedure CREATE|DROP INDEX syntax is ANSI standard syntax. You can specify the
UNIQUE option in the CREATE INDEX statement. You cannot use the NOMISS option with the
SQL procedure.

Managing Indexes with PROC SQL

Advantages Disadvantages

Additional indexes can be You can only create

created without re-creating indexes on existing SAS
the original indexes. data sets and existing
variables.
One or more indexes can The CREATE INDEX
be deleted without deleting statement cannot perform
all of the indexes on the data manipulation.
data set.
If an index exists, it must
be deleted before it can be
re-created.
58
3-26 Chapter 3 Accessing Observations

Comparing Techniques for Index Creation

INDEX= Data Set PROC DATASETS PROC SQL
Option
You can create the SAS data You can only create indexes You can only create indexes
set at the same time that the on existing SAS data sets on existing SAS data sets
index is created. and existing variables. and existing variables.

To create an additional index, Additional indexes can be Additional indexes can be

you must re-create the created without re-creating created without re-creating
existing indexes. the original indexes. the original indexes.

The DATA step can perform PROC DATASETS cannot The CREATE INDEX
data manipulation at the same perform data manipulation. statement cannot perform
time that the index is created. data manipulation.

To delete one or more One or more indexes can be One or more indexes can be
indexes, you must re-create deleted without deleting all of deleted without deleting all of
the other required indexes. the indexes on the data set. the indexes on the data set.

An existing index can be re- If an index exists, it must be If an index exists, it must be
created without first deleting it. deleted before it can be re- deleted before it can be re-
created. created.

59

Documenting Indexes
The following can be used to document indexes:
„ SAS Explorer

„ PROC CONTENTS

„ PROC DATASETS

„ SAS Management Console

60
 3.1 Creating an Index 3-27

Properties Window in SAS Explorer

61

Index Documentation
proc contents data=orion.sales_history; These
run; two
steps
proc datasets lib=orion nolist; produce
contents data=sales_history; identical
quit; output.

Partial PROC DATASETS Output

Alphabetic List of Indexes and Attributes

# of
Unique Unique
# Index Option Values Variables

1 Customer_ID 1046
2 Product_Group 56
3 SaleID YES 1500 Order_ID Product_ID

62 p303d04
3-28 Chapter 3 Accessing Observations

Exercises

Level 1

1. Creating Indexes
a. Open the program p303e01, and add the INDEX= option to create two indexes:
• a simple index Customer_ID, based on the variable Customer_ID
• a unique index Order_ID, based on the variable Order_ID
b. Use PROC SQL to delete the Order_ID index from the orders data set.
c. Use PROC DATASETS to create a composite index named OrDate based on the Order_ID and
Order_Date variables for the orders data set.
d. Use PROC CONTENTS or PROC DATASETS to look at the index information.

Level 2

2. Updating Indexes
a. Use the orion.price_list SAS data set to create a temporary data set named price_list that contains
a new variable named Unit_Profit that is the difference between the variables Unit_Sales_Price
and Unit_Cost_Price. Create a unique index on the Product_ID variable.
b. Open the program p303e02 and submit it.
c. View the log, and determine whether the new observation was added.
d. Why or why not?

Level 3

3. Creating Indexes on New Variables

a. Create a temporary SAS data set named all_staff by concatenating the data sets orion.sales and
orion.nonsales.
Hint: Rename the variables First and Last in orion.nonsales to be consistent with the variables
First_Name and Last_Name in orion.sales.
b. Create a new variable named Age_Hired that is the number of years between the variables
Hire_Date and Birth_Date.
c. Index the all_staff data set on the variable Age_Hired.
 3.2 Using an Index 3-29

3.2 Using an Index

Objectives
„ Describe when an index is used for WHERE statement
processing.
„ Describe when an index is not used for WHERE
statement processing.

66

Index Usage Possible

An index might be used when a WHERE expression
references one of the following:
„ a simple index key variable

„ the primary key variable of a composite index

Although a WHERE expression can consist of

multiple conditions that specify different variables,
SAS uses only one index to process the WHERE
expression.

67
3-30 Chapter 3 Accessing Observations

Index Usage Possible

A WHERE condition might possibly use an index,
provided the condition contains any one of the following:
„ a comparison operator or the IN operator

„ the NOT operator

„ the special WHERE operators (CONTAINS, LIKE,

IS NULL|IS MISSING, and BETWEEN…AND)
„ the TRIM or SUBSTR functions (if the second
argument of the SUBSTR function is 1)

68

General form of the SUBSTR function:

SUBSTR(variable, position, <length>);

Subtle improvements were made to the circumstances under which SAS uses an index in SAS 9.2.
Trailing blanks in the CONTAINS operator pattern to be searched for are ignored. Escape characters in
the LIKE operator are permitted. Examples are provided in the table below:

Condition Examples

Comparison operators and the IN where Customer_ID=14864;

operator where Order_ID < 5000;
where Customer_ID in (9313,14864);

Comparison operators with NOT where Customer_ID ne 14864;

where Customer_ID not in (9313,14864);

Comparison operators with the colon where Product_Group =: 'S';

modifier
The colon modifier (=:) indicates a
starts with condition. It cannot be used
in the SQL procedure.

CONTAINS operator where Product_Group contains 'Eclipse';

where Product_Group contains 'Eclipse ';
 3.2 Using an Index 3-31

Fully bounded range conditions where 5000 < Order_ID < 10000;
specifying both an upper and a lower where Order_ID between 5000 and 10000;
limit, which includes the BETWEEN-
AND operator

Pattern-matching operator LIKE where Product_Group like '%Shoes';

where Product_Group like 'G___';
where Product_Group like 'Girls\_Shoes';

IS NULL or IS MISSING operator where Product_Group is null;

where Product_Group is missing;

TRIM function where trim(Product_Group)='Clothes';

The SUBSTR function with the where substr(Product_Group,1,5)= 'Orion';

conditions that the starting position=1
and the length is less than or equal to
the length of the string variable

 For more information about when index usage is possible, see SAS 9.2 Language Reference:
Concepts Ö SAS Files Concepts Ö SAS Data Files Ö Understanding SAS Indexes in the
Help facility.

Setup for the Poll

The following indexes were created on the
orion.sales_history data set.
Partial PROC DATASETS Output
Alphabetic List of Indexes and Attributes

# of
Unique Unique
# Index Option Values Variables

1 Customer_ID 1046
2 Product_Group 56
3 SaleID YES 1500 Order_ID Product_ID

70
3-32 Chapter 3 Accessing Observations

3.05 Multiple Answer Poll

Which of the following WHERE conditions could possibly
use an index?
a. where Product_ID=220100300042;
b. where Customer_ID ne 3245;
c. where Customer_ID=15020 or
Customer_ID=14853;
d. where Order_ID=1230036183;
e. where Customer_ID='3245';

71

When Is an Index Not Used?

An index is not used in the following circumstances:
„ with a subsetting IF statement in a DATA step

„ with particular WHERE expressions

„ if SAS determines that all observations will satisfy

the WHERE expression
„ if SAS determines that it is more efficient to read
the data sequentially

73
 3.2 Using an Index 3-33

3.06 Multiple Choice Poll

When does the subsetting IF statement select
observations?
a. before the observation is copied into the PDV
b. after the observation is in the PDV

75

Using a Subsetting IF

Input
SAS Buffers
The subsetting IF
Data statement
selects observations.

PDV
ID Gender Country Name
Output Buffers
SAS
Data

77
3-34 Chapter 3 Accessing Observations

No Index Usage
SAS does not use an index when a WHERE expression
references an indexed variable if the following conditions
exist:
„ No single index can supply all required observations.

„ Any function other than TRIM or SUBSTR appears

in the WHERE expression.
„ The SUBSTR function does not search a string
beginning at the first position.
„ The SOUNDS-LIKE operator (=*) is used.

78

Condition Examples

No single index can supply all where Employee_ID=99999999 or Order_Type=1;

required observations.

Any function other than TRIM or where scan(Product_Group,2,' ')='Shoes';

SUBSTR appears in the WHERE
expression.

The SUBSTR function does not where substr(Product_Group,5,1)=' ';

search a string beginning at the first
position.

The SOUNDS-LIKE operator (=*) is where Product_Group=*'gulf';

used.

 For more information about when an index is not used, see SAS 9.2 Language Reference:
Concepts Ö SAS Files Concepts Ö SAS Data Files Ö Understanding SAS Indexes in the
Help facility.
 3.2 Using an Index 3-35

Compound Optimization
A WHERE expression that references multiple variables
can take advantage of a composite index.

compound use of a composite index to optimize

optimization some WHERE expressions that involve
multiple variables

where Order_ID=240200100038 and

Product_ID=1230151326;

79

Compound Optimization
For compound optimization to occur, all of the following
must be true:
„ At least the first two key variables in the composite
index must be used in the WHERE conditions.
„ The conditions must be connected using the AND
operator.
„ At least one condition must use the EQ, equal sign (=),
or IN operator.

80
3-36 Chapter 3 Accessing Observations

3.07 Multiple Choice Poll

Which of the following WHERE statements can use the
composite index SaleID for compound optimization?
a. where Order_ID=240200100038 or
Product_ID=1230151326;
b. where Order_ID=. and
Product_ID=1230151326;
c. where int(Order_ID/1000000000)=240
and Product_ID=1230151326;
d. where Order_ID>240000000000 and
Product_ID<1240000000;

82

WHERE Expression Index Usage

SAS uses the following steps to decide whether
to evaluate a WHERE expression using a sequential
read or using an index:
„ Determine whether the WHERE expression can
be satisfied by an existing index.
„ Select the best index, if several indexes are available.

„ Estimate the number of observations that qualify.

„ Compare the probable resource usage for both

methods.

 SAS estimates the I/O operations for indexed

access based on the subset size and sort order.

84
 3.2 Using an Index 3-37

Subset Size
SAS might
use an index.

SAS will
probably
33.3% use an index.

3%
0%
Data Set
SAS will
use an index.

87

To determine whether it is more efficient to satisfy the WHERE expression by using the index or by
reading the data sequentially, SAS uses these guidelines:
• If only a few observations are qualified, it is more efficient to use the index than to do a sequential
search of the entire data file.
• If most or all of the observations qualify, then it is more efficient to read the data file sequentially.

 If the subset is between small and large, other factors such as data order are important.
3-38 Chapter 3 Accessing Observations

Subset Size
The SAS index includes cumulative percentiles or
centiles. By default, SAS stores 21 centiles or every
5th percentile of the index. This information is used
to estimate the size of a qualifying subset.

centiles provide information about the distribution

of values in an index.

88

 For information about updating and viewing the centile information, see the UPDATE
CENTILES information in the SAS documentation for the DATASETS procedure and the
CENTILES option for the PROC CONTENTS statement.

3.08 Multiple Choice Poll

Which of the following is used to determine the I/O to read
a SAS data set sequentially?
a. the page size of the input data set and the number
of buffers available
b. the number of observations and the number
of variables
c. the page size of the output data set and the number
of output buffers available

90
 3.2 Using an Index 3-39

Review of Factors Affecting I/O

The following factors affect I/O:
„ size of the subset relative to the size of the data file

„ order of data with regard to the chosen index

„ page size of the data file

„ number of buffers allocated

„ cost to uncompress a compressed file for a sequential

read

92

Data Order
Obs Customer_ID
. For data that is sorted
.

8939
.
56487
and indexed on the same
8940
8941
70175
74667
variable(s), retrieval time
.
. through the index is much
.
faster than either sorted or
.
.
indexed data alone.
.
32548 89619 where Customer_ID in
32549 70187 (70201, 70187, 70175);
32550 76278
.
.
.

.
Fewer pages are
.
.
copied into memory
45775 84989 if the data is sorted.
45776 70201
45777 20209
.
.
.
Unsorted data Sorted data
93

 All of the observations meeting a specific criteria (Customer_ID = 14844) are on the same or
adjacent data set pages. Thus, fewer data set pages must be read to retrieve the same selected
observations.
3-40 Chapter 3 Accessing Observations

Controlling WHERE Processing Index Usage

You can control index usage for WHERE processing
with these data set options:
IDXWHERE=YES tells SAS to choose the best index to
optimize a WHERE expression and to
disregard the possibility that a sequential
search of the data set might be more
resource efficient.
IDXWHERE=NO tells SAS to ignore all indexes and satisfy
the conditions of a WHERE expression
with a sequential search of the data set.
IDXNAME=index-name directs SAS to use a specific index.

Use the IDXWHERE=NO option when you know

an available index will not optimize WHERE clause
processing.
94

Using the IDXWHERE= Option

To ensure that SAS uses an index when printing the data
for Customer_ID in (14844,4983,5862,10032) and
Product_Group contains 'Shoes', use the following
code:

options msglevel=i;
proc print data=orion.sales_history(idxwhere=yes);
where Customer_ID in (14844,4983,5862,10032)
and Product_Group contains 'Shoes';
var Customer_ID Product_ID Product_Group ;
title 'With an Index';
run;

p303d05
95
 3.2 Using an Index 3-41

Using the IDXWHERE= Option

Partial SAS Log
1669 options msglevel=i;
1670 proc print data=orion.sales_history(idxwhere=yes);
1671 where Customer_ID in (14844,4983,5862,10032)
1672 and Product_Group contains 'Shoes';
INFO: Data set option (IDXWHERE=YES)forced an index to be used rather
than a sequential pass for where-clause processing.
INFO: Index Customer_ID selected for WHERE clause optimization.
1673 var Customer_ID Product_ID Product_Group ;
1674 title 'With an Index';
1675 run;

p303d05
96

Using the IDXNAME= Option

Because using the index on Customer_ID returns
a smaller subset than would the index on
Product_Group, the IDXNAME= data set option can be
used.
options msglevel=i;
proc print data=orion.sales_history(idxname=Customer_ID);
where Customer_ID in (14844,4983,5862,10032)
and Product_Group contains 'Shoes';
var Customer_ID Product_ID Product_Group ;
title 'With an Index';
run;

Use the IDXNAME= option when you know the

better index so SAS does not need to do the
evaluation.
p303d06
97
3-42 Chapter 3 Accessing Observations

Using the IDXNAME= Option

Partial SAS Log
92 options msglevel=i;
193 proc print data=orion.sales_history(idxname=Customer_ID);
194 where Customer_ID in (14844,4983,5862,10032)
195 and Product_Group contains 'Shoes';
INFO: Index Customer_ID selected for WHERE clause optimization.
196 var Customer_ID Product_ID Product_Group ;
197 title 'With an Index';
198 run;

NOTE: There were 3 observations read from the data set

ORION.SALES_HISTORY.
WHERE Customer_ID in (4983, 5862, 10032, 14844) and
Product_Group contains 'Shoes';
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.15 seconds
cpu time 0.01 seconds

98
 3.2 Using an Index 3-43

Maintaining Indexes
Data Management Tasks Index Action Taken
Copy the data set with the Index file constructed for
COPY procedure or the new data file
DATASETS procedure
Move the data set with the Index file deleted from IN=
MOVE option in the COPY library; rebuilt in OUT=
procedure library
Copy the data set with a Index file constructed for
drag-and-drop action in new file
SAS Explorer

99 continued...

Maintaining Indexes
Data Management Tasks Index Action Taken
Rename the data set Index file renamed
Rename the variable Variable renamed to new
name in index file
Add observations Value/Identifier pairs added
Delete observations Value/Identifier pairs
deleted; space recovered
for re-use
Update observations Value/Identifier pairs
updated if values change
 The APPEND procedure and the INSERT INTO
statement in the SQL procedure update the index file
after all the data is appended or inserted.
100 continued...

Indexes are maintained by updates in place, such as using the VIEWTABLE window to update, add, or
delete observations, and the APPEND or SQL procedure to append data. Using the Explorer window or
the DATASETS procedure also maintains indexes when data sets or variables are renamed. However,
re-creating a data set with the SET, MERGE, or UPDATE statement does not automatically maintain
indexes.
3-44 Chapter 3 Accessing Observations

Maintaining Indexes
Data Management Tasks Index Action
Taken
Delete a data set. Index file deleted
proc datasets lib=work;
delete a;
run;
Rebuild a data set with a DATA step or the Index file deleted
SQL procedure.
data a; proc sql;
set a; create table a as
run; select * from a;
quit;
Sort the data set in place with the FORCE Index file deleted
option in the SORT procedure.
proc sort data=a force;
by var;
run;
101

If you use the UPLOAD procedure or the DOWNLOAD procedure in SAS/CONNECT, the index is
re-created by default when you upload or download a single data set and omit the OUT= option or when
you upload or download a SAS data library. Use the INDEX=NO data set option to upload or download
without re-creating the index.
Index re-created:
proc upload data=schedule;
run;
Index not re-created:
proc download data=Sales(index=no);
run;
If you are using the CPORT procedure to create transport files, you can use the INDEX=YES option in
the PROC CPORT statement to transport the index file along with the data set. INDEX=YES is the
default.
 3.2 Using an Index 3-45

Guidelines for Indexing

Suggested guidelines for creating indexes:
„ Create an index when you intend to retrieve a small
subset of observations from a large data file.
„ Do not create an index if the data file page count is
less than three pages. It is faster to access the data
sequentially.
„ Create indexes on variables that are discriminating.
These variables precisely identify observations that
satisfy WHERE expressions.
„ When you create a composite index, make the first key
variable the most discriminating.
„ Consider the cost of maintaining an index for a data file
that is frequently changed.
102 continued...

 A variable such as Gender is not discriminating. A discriminating variable is one that enables
you to break the data into many small groups or subsets.

Guidelines for Indexing

„ To minimize I/O for indexed access, sort the data by
the key variable(s) before creating the index. Maintain
the data file in sorted order by the key variable to
improve performance.
„ Minimize the number of indexes to reduce disk storage
and update costs. Create indexes only on variables
that are often used in queries or BY-group processing
(when the data cannot be sorted).
„ Consider how often your applications use an index.
An index must be used often in order to compensate
for the resources used in creating and maintaining it.
„ When you create an index to process a WHERE
expression, do not try to create one index that might
be used to satisfy every conceivable query.
103
3-46 Chapter 3 Accessing Observations

Index Trade-offs
Advantages Disadvantages

fast access to a small extra CPU cycles and I/O

subset of observations operations to create and
maintain an index
values returned in sorted increased CPU to read
order the data
can enforce uniqueness extra disk space to store
the index file
extra memory to load the
index pages and the
compiled SAS C code to
use the index
104
 3.2 Using an Index 3-47

Exercises

Level 1

4. Using an Index
Open the program p303e04, and submit it. Consult the log and answer the questions following the
program code shown here.
p303e04
options msglevel=I;
*** Example 1;

data rdu;
set orion.sales_history;
if Order_ID=1230166613;
run;

*** Example 2;

proc print data=orion.sales_history;

where Order_ID=1230166613 or Product_ID=220200100100;
run;

*** Example 3;

proc print data=orion.sales_history;

where Product_Group ne 'Shoes';
run;

*** Example 4;

proc print data=orion.sales_history;

where Customer_ID=12727;
run;

**** Example 5;

proc print data=orion.sales_history;

where Product_ID=220200100100;
run;

*****Example 6;

data saleshistorycopy;
set orion.sales_history;
run;
3-48 Chapter 3 Accessing Observations

Questions:
a. Does Example 1 use an index? Why or why not?

Replace the IF statement with a WHERE statement, and resubmit the program. Does the example
now use an index? Why or why not?

b. Does Example 2 use an index? Why or why not?

Replace the OR operator with the AND operator, and resubmit the program. Does the example
now use an index? Why or why not?

c. Does Example 3 use an index? Why or why not?

Replace the NE operator with the EQ operator, and resubmit the program. Does the example now
use an index? Why or why not?

d. Does Example 4 use an index? Why or why not?

Add the IDXWHERE=NO data set option and resubmit the program. Is the output from the
PROC PRINT step with an index different from the output from the PROC PRINT step without
an index?
What message do you see in the log?
 3.2 Using an Index 3-49

e. Does Example 5 use an index? Why or why not?

f. In Example 6, does the data set saleshistorycopy have an index?

Level 2

5. Suppressing Index Usage

Create a detail report from the orion.supplier SAS data set that lists all of the variables and
observations where Supplier_ID is greater than 1000. Ensure that the data is processed sequentially.

Level 3

6. Updating Centile Information for an Index

a. Submit program p303e06. Do not clear your SAS Output window or the Results window!
p303e06
data orders(index=(Order_ID/unique Customer_ID));
set orion.orders;
run;
proc contents data=orders centiles;
run;
b. Using the DATASETS procedure, set the indicator for updating the centile information about the
Order_ID index to 1% of the data.
c. Submit the program p303e06c, which adds new observations to the orders data set.
d. Submit a PROC CONTENTS step to view the contents of orders. Compare the centile
information from step 6.a. to the current centile information. Were the centiles updated or not?
______________________________________________________________________________
e. Why or why not?
______________________________________________________________________________
______________________________________________________________________________
3-50 Chapter 3 Accessing Observations

3.3 Creating a Sample Data Set (Self-Study)

Objectives
„ Create a systematic sample.
„ Create a random sample with replacement.
„ Create a random sample without replacement.

108

Business Scenario
The Marketing Department wants to send customer
satisfaction questionnaires to a sample of the customers
in the orion.order_fact SAS data set.
Partial Listing of orion.order_fact
Customer Employee Delivery_
Street_ID Order_Date Order_ID . . .
_ID _ID Date
63 121039 9260125492 11JAN2003 11JAN2003 1230058123 ...
5 99999999 9260114570 15JAN2003 19JAN2003 1230080101 ...
45 99999999 9260104847 20JAN2003 22JAN2003 1230106883 ...
41 120174 1600101527 28JAN2003 28JAN2003 1230147441 ...
183 120134 1600100760 27FEB2003 27FEB2003 1230315085 ...
. . . . . .
. . . . . .
. . . . . .

109
 3.3 Creating a Sample Data Set (Self-Study) 3-51

Business Scenario
Select a subset by reading every 50th observation from
observation number 1 to the end of the SAS data set.

data subset;
e do PickIt=1 to TotObs by 50; d
set orion.order_fact(keep=Customer_ID
Employee_ID Street_ID Order_ID)
point=PickIt
nobs=TotObs; c
output; f
end;
stop; g
run;

p303d07
110

c The NOBS= option creates a temporary numeric variable that contains the total number of
observations in the input data. This variable is populated at compilation.
d You can refer to the NOBS= variable in executable statements that appear before the SET statement.
e The DO loop assigns a value to the variable PickIt. PickIt is used by the POINT= option in the SET
statement to select an observation from the SAS data set. PickIt must have a value before the SET
statement executes.
f The OUTPUT statement writes the PDV values to the SAS data set.
g The STOP statement stops the DATA step from continuing to execute after the five observations are
selected. Without a STOP statement, the DATA step continues in an infinite loop.
3-52 Chapter 3 Accessing Observations

3.09 Quiz
Are POINT= and NOBS= individual statements
or part of the SET statement?
data subset;
do PickIt=1 to TotObs by 50;
set orion.order_fact(keep=Customer_ID
Employee_ID Street_ID Order_ID)
point=PickIt
nobs=TotObs;
output;
end;
stop;
run;

p303d07
112

Using the POINT= Option

To create a sample, use the POINT= option in the
SET statement.
General form of the POINT= option:
SET data-set-name POINT=point-variable;

The point-variable has the following attributes:

„ names a temporary numeric variable that contains
the number of the observation to read
„ must be given a value before the execution of the
SET statement
„ must be a variable (for example, X) and not a constant
value (for example, 12)
„ must be a valid observation number

114

The POINT= option value should be an integer greater than zero and less than or equal to the number of
observations in the SAS data set.
• If the value is not integral, the SET statement effectively applies the FLOOR function to the value.
• If, during processing, the POINT= value does not match an observation number (is negative or is
greater than NOBS), a data error results and no observation is read by the SET statement. The DATA
step will output the current contents of the PDV and continue processing.
 3.3 Creating a Sample Data Set (Self-Study) 3-53

Using the Number of Observations

You can use the NOBS= option in the SET statement
to determine how many observations there are in a
SAS data set.
General form of the SET statement:

SET SAS-data-set NOBS=variable;

The NOBS= option creates a temporary variable whose

value has the following characteristics:
„ is the number of observations in the input data set(s)

„ is assigned during compilation

„ is retained

„ should not be modified during execution

115

Using the STOP Statement

The POINT= option has the following features:
„ uses direct-access read mode

„ does not detect the end-of-file marker

To prevent the DATA step from looping continuously,

use the STOP statement.
General form of the STOP statement:

STOP;

116
3-54 Chapter 3 Accessing Observations

Compilation
data subset;
do PickIt=1 to TotObs by 50;
set orion.order_fact
(keep=Customer_ID
Employee_ID
Street_ID
Order_ID)
point=PickIt
nobs=TotObs;
output;
end;
stop;
run;
PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
. 617 . . . . .

p303d07
117 ...

During compilation, the value for TotObs is retrieved from the descriptor portion of orion.order_fact.

Execution
Partial Listing of data subset;
orion.order_fact do PickIt=1 to TotObs by 50;
Customer Employee
. . .
set orion.order_fact
obs _ID _ID (keep=Customer_ID
1 63 121039 . . . Employee_ID
2 5 99999999 . . . Street_ID
. .
Order_ID)
. . . . . point=PickIt
. . nobs=TotObs;
50 17023 99999999 . . . output;
51 17023 99999999 . . . end;
. .
stop;
. . . . . run;
. .

PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
1 617 63 121039 9260125492 1230058123 1

119 ...

The SET statement executes and reads the first observation. The first observation is read because the
variable PickIt has a value of 1, not because SAS is reading sequentially.
 3.3 Creating a Sample Data Set (Self-Study) 3-55

Execution
Partial Listing of data subset;
orion.order_fact do PickIt=1 to TotObs by 50;
Customer Employee
. . .
set orion.order_fact
obs _ID _ID (keep=Customer_ID
1 63 121039 . . . Employee_ID
2 5 99999999 . . . Street_ID
. .
Order_ID)
. . . . . point=PickIt
. . nobs=TotObs;
50 17023 99999999 . . . output;
51 17023 99999999 . . . end; Output current
. .
stop; observation.
. . . . . run;
. .

PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
1 617 63 121039 9260125492 1230058123 1

120 ...

Execution
Partial Listing of data subset;
orion.order_fact do PickIt=1 to TotObs by 50;
Customer Employee
. . .
set orion.order_fact
obs _ID _ID (keep=Customer_ID
1 63 121039 . . . Employee_ID
2 5 99999999 . . . Street_ID
. .
Order_ID)
. . . . . point=PickIt
. . nobs=TotObs;
50 17023 99999999 . . . output;
51 17023 99999999 . . . end;
. .
stop;
. . . . . run;
. .

PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
51 617 17023 99999999 2600100021 1230931366 1

123 ...

This time when the SET statement executes, observation 51 is read from orion.order_fact.
3-56 Chapter 3 Accessing Observations

Execution
Partial Listing of data subset;
orion.order_fact do PickIt=1 to TotObs by 50;
Customer Employee
. . .
set orion.order_fact
obs _ID _ID (keep=Customer_ID
1 63 121039 . . . Employee_ID
2 5 99999999 . . . Street_ID
. .
Order_ID)
. . . . . point=PickIt
. . nobs=TotObs;
50 17023 99999999 . . . output;
51 17023 99999999 . . . end; Output current
. .
stop; observation.
. . . . . run;
. .

PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
51 617 17023 99999999 2600100021 1230931366 1

124 ...

Execution PickIt > TotObs

Partial Listing of data subset;
orion.order_fact do PickIt=1 to TotObs by 50;
Customer Employee
. . .
set orion.order_fact
obs _ID _ID (keep=Customer_ID
1 63 121039 . . . Employee_ID
2 5 99999999 . . . Street_ID
. .
Order_ID)
. . . . . point=PickIt
. . nobs=TotObs;
50 17023 99999999 . . . output;
51 17023 99999999 . . . end;
. .
stop;
. . . . . run;
. .

PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
651 617 215 120175 1600102721 1243963366 1

126 ...

When PickIt has a value of 651, its value is greater than the range (1-617) in the iterative DO loop.
 3.3 Creating a Sample Data Set (Self-Study) 3-57

Execution
Partial Listing of data subset;
orion.order_fact do PickIt=1 to TotObs by 50;
Customer Employee
. . .
set orion.order_fact
obs _ID _ID (keep=Customer_ID
1 63 121039 . . . Employee_ID
2 5 99999999 . . . Street_ID
. .
Order_ID)
. . . . . point=PickIt
. . nobs=TotObs;
50 17023 99999999 . . . output;
51 17023 99999999 . . . end;
. .
stop; Execution stops.
. . . . . run;
. .

PDV
Tot Customer_ Employee_ Street_
D PickIt D Order_ID D _N_
Obs ID ID ID
651 617 215 120175 1600102721 1243963366 1

127 ...

Control goes to the next executable statement after the end of the DO loop.

Resulting Data Set

Partial Listing of subset
Systematic Sample

Obs Customer_ID Employee_ID Street_ID Order_ID

1 63 121039 9260125492 1230058123

2 17023 99999999 2600100021 1230931366
3 17 121037 9260123306 1231757107
4 195 120160 1600101663 1232590052
5 41 120134 1600101527 1233545775
6 11171 99999999 2600100032 1235176942
7 10 121043 9260129395 1237327705
8 53 120121 1600103258 1238674844
9 90 121040 9260111614 1239543223
10 89 121061 9260116551 1240549230
11 27 99999999 9260105670 1241930625
12 41 120195 1600101527 1242838815
13 215 120175 1600102721 1243963366

128
3-58 Chapter 3 Accessing Observations

Creating a Random Sample

Instead of creating a systematic sample, create a random
sample where each observation has an equal chance of
being selected.
There are two types of random samples:
„ with replacement, where an observation might be
selected more than one time
„ without replacement, where an observation cannot
be selected more than once
You can use the RANUNI function to generate random
numbers from a uniform distribution.
General form of the RANUNI function:

RANUNI(seed)

129

The UNIFORM function is an alias for the RANUNI function.

The seed is an initial starting point that the RANUNI function uses to generate streams of random
numbers. The seed must be an integer with a value less than 231-1 (2,147,483,647).

 A 0 argument for the RANUNI function uses the system clock time, which results in a different
stream of random numbers each time that the program is run.
 3.3 Creating a Sample Data Set (Self-Study) 3-59

Using the RANUNI Function

The RANUNI function returns a rational number
between 0 and 1 (non-inclusive) generated from a
uniform distribution.

0 1
ranuni(seed)

Examples:
Random number
.01253689
.95196500

130 ...

Using the RANUNI Function

If you want a number between 0 and 5 (non-inclusive),
multiply the number returned from the RANUNI function
by 5.

0 5
ranuni(seed) * 5

Examples:
Random number * 5
.01253689 Î 0.06268445
.95196500 Î 4.75982500

131 ...
3-60 Chapter 3 Accessing Observations

Using the RANUNI and CEIL Functions

If you want an integer between 1 and 5 (inclusive), use
the CEIL function on the number returned by multiplying
the random number by 5.

1 2 3 4 5
ceil(ranuni(seed) * 5)

Examples:
Random number * 5 CEIL( )
.01253689 Î 0.06268445 Î 1
.95196500 Î 4.75982500 Î 5

132

The CEIL function returns the smallest integer that is greater than or equal to the argument.

Setup for the Poll

Instead of the CEIL function, would the INT function return
the same results?
ceil(ranuni(seed) * 5)

int(ranuni(seed) * 5)

134
 3.3 Creating a Sample Data Set (Self-Study) 3-61

3.10 Poll
Instead of the CEIL function, would the INT function return
the same results?
€ Yes
€ No

135
3-62 Chapter 3 Accessing Observations

Creating a Random Sample

p303d08
Create a random sample with replacement. A sample with replacement can contain duplicate observations
because an observation can be selected more than one time.
p303d08
data subset(drop=i SampSize);
SampSize=10;
do i=1 to SampSize;
PickIt=ceil(ranuni(0)*TotObs);
ObsPicked=PickIt;
set orion.order_fact point=PickIt nobs=TotObs;
output;
end;
stop;
run;

proc print data=subset;

title 'A Random Sample with Replacement';
var ObsPicked Customer_ID Order_Date Delivery_Date Order_ID;
run;
PROC PRINT Output
A Random Sample with Replacement

Obs Order_ Delivery_

Obs Picked Customer_ID Date Date Order_ID

1 17 20 01APR2003 01APR2003 1230498538

2 499 46966 07APR2007 08APR2007 1241909303
3 200 41 20AUG2004 20AUG2004 1233545775
4 416 12 03AUG2006 03AUG2006 1239836937
5 216 19873 27OCT2004 03NOV2004 1233998114
6 548 27 12JUL2007 17JUL2007 1242782701
7 290 70221 12AUG2005 14AUG2005 1236694462
8 46 75 03JUN2003 03JUN2003 1230841466
9 422 90 23AUG2006 23AUG2006 1239994933
10 129 53 13JAN2004 13JAN2004 1232087464

 With a seed value of 0, you get different results each time that the program is executed, but it is
possible that some of the same observations that were selected in previous executions will be
selected.
 3.3 Creating a Sample Data Set (Self-Study) 3-63

p303d09
Create a random sample without replacement. A sample without replacement cannot contain duplicate
observations because after an observation is output to work.subset, it cannot be selected again
programmatically.
p303d09
data subset(drop=ObsLeft SampSize);
c SampSize=10;
d ObsLeft=TotObs;
do while(SampSize>0 and ObsLeft>0);
e PickIt+1;
if ranuni(0)<SampSize/ObsLeft then
do;
ObsPicked=PickIt;
set orion.order_fact point=PickIt
nobs=TotObs;
output;
SampSize=SampSize-1;
end;
ObsLeft=ObsLeft-1;
end;
stop;
run;

proc print data=subset;

title 'A Random Sample without Replacement';
var ObsPicked Customer_ID Order_Date Delivery_Date Order_ID;
run;
c SampSize is the number of observations wanted in the sample.
d ObsLeft is the number of observations that still need to be selected. The start value is equal to
TotObs, the total number of observations in the data set being sampled.
e PickIt is the number of the observation to be read in the sample data set. Because it is used in a
SUM statement, its starting value is 0.
3-64 Chapter 3 Accessing Observations

PROC PRINT Output

A Random Sample without Replacement

Obs Order_ Delivery_

Obs Picked Customer_ID Date Date Order_ID

1 48 56 11JUN2003 11JUN2003 1230885738

2 50 17023 20JUN2003 25JUN2003 1230931366
3 51 17023 20JUN2003 25JUN2003 1230931366
4 128 45 31DEC2003 31DEC2003 1232007700
5 153 89 31MAR2004 03APR2004 1232601472
6 162 34 16APR2004 16APR2004 1232709115
7 276 36 15JUN2005 18JUN2005 1236113431
8 388 45 29MAY2006 29MAY2006 1239312711
9 410 13 23JUL2006 28JUL2006 1239744161
10 586 171 16OCT2007 16OCT2007 1243643970

 With a seed value of 0, you get different results each time that the program is executed, but it is
possible that some of the same observations will be selected as were selected in previous
executions.
In each iteration of the DO loop, the following occur:
1. PickIt is incremented by 1.
2. The IF expression ranuni(0) < Sampsize/ObsLeft is evaluated.
a. If true, these actions occur:
1) The observation PickIt is selected in the sample.
2) SampSize is decreased by 1.
b. If false, the observation PickIt is skipped.
3. ObsLeft is decreased by 1.
The process ends when SampSize is 0; no additional observations are needed.
Be aware of the following:
• Each observation is considered for selection.
• An observation number is considered only once.
• The data set is read only when an observation number is selected.

 This is an adaptation of a sampling routine that was used by statisticians for many years.
• The sample size is fixed.
• An observation can be selected only once.
• Each observation has an equal probability of being selected.
• The selection probability for an observation is independent of the selection of another
observation.
 3.3 Creating a Sample Data Set (Self-Study) 3-65

Using the SURVEYSELECT Procedure

The SURVEYSELECT procedure has the following
attributes:
„ provides a variety of methods for selecting probability-
based random samples
„ can select a simple random sample or can sample
according to a complex multistage sample design
that includes stratification, clustering, and unequal
probabilities of selection
„ is part of SAS/STAT

138

Using the SURVEYSELECT Procedure

This program creates a SAS data set, ordersample, that
contains 10 observations randomly selected, without
replacement, from the orion.order_fact SAS data set.

proc surveyselect data=orion.order_fact

(keep=Customer_ID Employee_ID
Street_ID Order_ID)
out=ordersample
method=srs n=10;
run;

p303d10
139
3-66 Chapter 3 Accessing Observations

Using the SURVEYSELECT Procedure

General form of the SURVEYSELECT procedure:

PROC SURVEYSELECT options;

STRATA variables;
CONTROL variables;
SIZE variable;
ID variables;
RUN;

140

STRATA partitions the input data set into non-overlapping groups defined by the
STRATA variables. PROC SURVEYSELECT then selects independent
samples from these strata, according to the selection method and design
parameters specified in the PROC SURVEYSELECT statement. PROC
SURVEYSELECT expects the input data set to be sorted in the order of the
STRATA variables.

CONTROL names variables for sorting the input data set. The CONTROL variables
can be character or numeric. PROC SURVEYSELECT sorts the input data
set by the CONTROL variables before selecting the sample. If you also
specify a STRATA statement, PROC SURVEYSELECT sorts by the
CONTROL variables within the strata.

SIZE names one and only one size measure variable, which contains the size
measures to be used when sampling with probability proportional to size.
The SIZE variable must be numeric. When the value of an observation's
SIZE variable is missing or non-positive, that observation has no chance of
being selected for the sample.

ID names variables from the DATA= input data set to be included in the
OUT= data set of selected units. If there is no ID statement, PROC
SURVEYSELECT includes all variables from the DATA= data set in the
OUT= data set. The ID variables can be character or numeric.
 3.3 Creating a Sample Data Set (Self-Study) 3-67

Using the SURVEYSELECT Procedure

The PROC SURVEYSELECT statement performs
the following tasks:
„ invokes the procedure

„ can, if you choose, identify input and output data sets

„ specifies the sample selection method, the sample

size, and other sample design parameters
The PROC SURVEYSELECT statement is the only
statement required to create a simple random sample.

141

Options for the SURVEYSELECT Procedure

The following options can be specified in the
PROC SURVEYSELECT statement:

To do this: Use this option:

Specify the input data set DATA=
Specify the output data set OUT=
Suppress displayed output NOPRINT
Specify selection method METHOD=
Specify sample size SAMPSIZE=
N=
Specify random number seed SEED=

142
3-68 Chapter 3 Accessing Observations

Methods Used by the SURVEYSELECT

Procedure
Selected values for the METHOD= option are as follows:
SYS This method of systematic random sampling
selects units at a fixed interval throughout the
sampling frame or stratum after a random start.

URS This method of unrestricted random sampling

selects units with equal probability and with
replacement. Because units are selected with
replacement, a unit can be selected for the
sample more than once.

SRS This method of simple random sampling selects

units with equal probability and without
replacement. The selection probability for each
individual unit equals n/N.

143

Using the SURVEYSELECT Procedure

This program creates a SAS data set, ordersample, that
contains 10 observations randomly selected, without
replacement, from the orion.order_fact SAS data set.

proc surveyselect data=orion.order_fact

(keep=Customer_ID Employee_ID
Street_ID Order_ID)
out=ordersample
method=srs n=10;
run;

p303d10
144
 3.3 Creating a Sample Data Set (Self-Study) 3-69

Using the SURVEYSELECT Procedure

In addition to creating the SAS data set, ordersample,
PROC SURVEYSELECT provides the following
information in the Output window:
The SURVEYSELECT Procedure

Selection Method Simple Random Sampling

Input Data Set ORDER_FACT

Random Number Seed 525990001 n
Sample Size 10
Selection Probability 0.016207 o
Sampling Weight 61.7 p
Output Data Set ORDERSAMPLE

145

c Because the SEED= option is not specified in the PROC SURVEYSELECT statement, the seed
value is obtained using the datetime value from the computer's clock.
d The Selection Probability for each individual unit is calculated as 10/617 (sample size/number of
observations in the input data set).
e The Sampling Weight is the inverse of the selection probability, 617/10.

Comparison of the DATA Step and the

SURVEYSELECT Procedure
DATA Step PROC SURVEYSELECT
full power of DATA step less coding
processing
can create multiple output one output data set with
data sets additional statistics
part of Base SAS part of SAS/STAT

146
3-70 Chapter 3 Accessing Observations

Exercises

Level 1

7. Generating a Systematic Sample

Generate a systematic sample by selecting every tenth supplier starting with observation 10 from the
data set orion.product_dim.
a. The sample should contain only the variables Product_Line, Product_ID, Product_Name, and
Supplier_Name.
b. Name the output data set products_sample.
c. Print the first five observations of products_sample. Omit the observation numbers.
Output
Systematic Sample of Products

Product_ID Product_Line Product_Name Supplier_Name

210200500002 Children Children's Mitten AllSeasons Outdoor Clothing

210200900033 Children Osprey France Nylon Shorts Triple Sportswear Inc
220100100044 Clothes & Shoes Sports glasses Satin Alumin. Eclipse Inc
220100100241 Clothes & Shoes Big Guy Men's Santos Shorts Dri Fit Eclipse Inc
220100100513 Clothes & Shoes Woman's Deception Dress Eclipse Inc

Level 2

8. Generating a Random Sample with Replacement

Generate a random sample with replacement of 50 customers from orion.customer_dim. If the
customer age is less than 40, place those customers in a data set named underforty. If the customer
age is greater than or equal to 40, place the customers in a data set named fortyplus.
 If you obtain zero observations in one of the data sets, run the program again. It is possible
that the selected observations might all be 40 or older or all could be less than 40. If you used
a constant seed in the RANUNI function, change the seed value and resubmit the program.

Level 3

9. Generating a Random Sample without Replacement

Generate a random sample without replacement of approximately 10% of the data in
orion.customer_dim.
 3.4 Chapter Review 3-71

3.4 Chapter Review

Chapter Review
1. What is one purpose of an index?

2. What are the three ways to create an index?

3. What SAS system option is used to view information

about index usage in the log?

4. How can you tell whether a SAS data set has an

index?

5. Which functions can use an index?

149 continued...

Chapter Review
6. Does a subsetting IF use an index?

7. Which DATA set option forces SAS to use an index for

WHERE clause processing?

8. Does sorting a data set before indexing help the index

perform better?

9. Does a DATA step using a SET statement that reads

and writes the same data set automatically maintain
an index?

150
3-72 Chapter 3 Accessing Observations

3.5 Solutions

Solutions to Exercises
1. Creating Indexes
a. Open the program p303e01, and add the INDEX= option to create two indexes:
• a simple index Customer_ID, based on the variable Customer_ID
• a unique index Order_ID, based on the variable Order_ID
p303s01
options msglevel=i;
data orders(index=(Customer_ID Order_ID / unique));
set orion.orders;
Days_To_Delivery=Delivery_Date - Order_Date;
run;
options msglevel=n;
b. Use PROC SQL to delete the Order_ID index from the orders data set.
proc sql;
drop index Order_ID
from orders;
quit;
c. Use PROC DATASETS to create a composite index OrDate based on the Order_ID and
Order_Date variables for the orders data set.
proc datasets library=work nolist;
modify orders;
index create OrDate=(Order_ID Order_Date);
quit;
d. Use PROC CONTENTS or PROC DATASETS to look at the index information.
/* CONTENTS solution */
proc contents data=orders;
run;
/* DATASETS solution */
proc datasets library=work nolist;
contents data=orders;
quit;
 3.5 Solutions 3-73

2. Updating Indexes
a. Use the orion.price_list SAS data set to create a temporary data set named price_list that
contains a new variable named Unit_Profit that is the difference of the variables
Unit_Sales_Price and Unit_Cost_Price. Create a unique index on the Product_ID variable.
p303s02
data price_list(index=(Product_ID / unique));
set orion.price_list;
Unit_Profit=Unit_Sales_Price - Unit_Cost_Price;
run;
b. Open the program p303e02 and submit it.
c. View the log and determine whether the new observation was added.
Partial SAS Log
208 /* Part b */
209 proc sql;
210 insert into price_list(Product_ID, Start_Date,
211 End_Date, Unit_Cost_Price,
212 Unit_Sales_Price, Factor,Unit_Profit)
213 values (210200100009, '15FEB2007'd, '31DEC9999'd, 15.50, 34.70, 1.00,
213! 19.20);
ERROR: Duplicate values not allowed on index Product_ID for file PRICE_LIST.
NOTE: This insert failed while attempting to add data from VALUES clause 1 to
the data set.
NOTE: Deleting the successful inserts before error noted above to restore table
to a consistent state.
214 quit;
NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE SQL used (Total process time):
real time 0.31 seconds
cpu time 0.04 seconds

d. Why or why not?

The new observation was not added because an observation with the value 210200100009
for Product_ID already exists in the data set price_list. The unique index prevents the
second observation from being added.
3. Creating Indexes on New Variables
a. Create a temporary SAS data set named all_staff by concatenating the data sets orion.sales and
orion.nonsales.
Hint: Rename the variables First and Last in orion.nonsales to be consistent with the variables
First_Name and Last_Name in orion.sales.
b. Create a new variable named Age_Hired that is the number of years between the variables
Hire_Date and Birth_Date.
3-74 Chapter 3 Accessing Observations

c. Index the all_staff data set on the variable Age_Hired.

p303s03
data all_staff(index=(Age_Hired));
set orion.sales orion.nonsales(rename=(First=First_Name
Last=Last_Name));
Age_Hired=int((Hire_Date - Birth_Date) / 365.25);
run;
4. Using an Index
Open the program p303e04, and submit it. Consult the log and answer the questions.
Questions:
a. Does Example 1 use an index? Why or why not?
No, the code uses an IF statement, not a WHERE statement.
Replace the IF statement with a WHERE statement, and resubmit the program. Does the example
now use an index? Why or why not?
Yes, the WHERE statement uses the SALEID index.

b. Does Example 2 use an index? Why or why not?

No, the WHERE statement uses the OR operator, which prohibits index use.
Replace the OR operator with the AND operator, and resubmit the program. Does the example
now use an index? Why or why not?
Yes, the WHERE statement uses both of the variables in the composite SALEID index with
the AND operator.

c. Does Example 3 use an index? Why or why not?

No, too much of the data satisfied the WHERE statement criteria.
Replace the NE operator with the EQ operator, and resubmit the program. Does the example now
use an index? Why or why not?
No, the data is too randomly distributed. The following INFO message is in the log:
INFO: Index Product_Group not used. Sorting into index order may help.
 3.5 Solutions 3-75

d. Does Example 4 use an index? Why or why not?

Yes, only a small subset of the data that satisfied the WHERE statement criteria was
returned.
Add the IDXWHERE=NO data set option and resubmit the program. Is the output from the
PROC PRINT step with an index different from the output from the PROC PRINT step without
an index?
No, the results are the same.
What message do you see in the log?
INFO: Data set option (IDXWHERE=NO) forced a sequential pass of the data rather than use
of an index for where-clause processing.

e. Does Example 5 use an index? Why or why not?

No, there is no index on the Product_ID variable in orion.sales_history. Even though
Product_ID is part of the composite Sale_ID index, it cannot be used because it is not the
primary key.
f. In Example 6, does the data set SalesHistoryCopy have an index?
No, the DATA step re-creates orion.sales_history as work.saleshistorycopy. The index is not
copied when the data set is created in the DATA step.
5. Suppressing Index Usage
Create a detail report from the orion.supplier SAS data set that lists all of the variables and
observations where Supplier_ID is greater than 1000. Ensure that the data is processed sequentially.
p303s05
options msglevel=i;
proc print data=orion.supplier(idxwhere=no);
where Supplier_ID > 1000;
run;
3-76 Chapter 3 Accessing Observations

6. Updating Centile Information for an Index

a. Submit the program p303e06. Do not clear your SAS Output window or the Results window!
Partial PROC CONTENTS Output
Alphabetic List of Indexes and Attributes

Current # of
Unique Update Update Unique
# Index Option Centiles Percent Values Variables

... Lines Removed ...

2 Order_ID YES 5 0 490

1230058123
1230699509
1231135703
1231544990
1231956902
1232601472
1233078086
1233920786
1234588648
1236055696
1237478988
1238353296
1238846184
1239408849
1240137702
1240692950
1241652707
1242265757
1242923327
1243568955
1244296274

b. Using the DATASETS procedure, set the indicator for updating the centile information about the
Order_ID index to 1% of the data.
p303s06
proc datasets lib=work nolist;
modify orders;
index centiles Order_ID / updatecentiles=1;
quit;
c. Submit the program p303e06c, which adds new observations to the work.orders data set.
 3.5 Solutions 3-77

d. Submit a PROC CONTENTS step to view the contents of orders. Compare the centile
information from step 6.a. to the current centile information. Were the centiles updated or not?
Yes
proc contents data=orders centiles;
run;
Partial PROC CONTENTS Output
Alphabetic List of Indexes and Attributes

Current # of
Unique Update Update Unique
# Index Option Centiles Percent Values Variables

... Lines Removed ...

2 Order_ID YES 1 0 495

1230058123
1230699509
1231169108
1231544990
1231976710
1232618023
1233131266
1233920795
1234665265
1236113431
1237664026
1238370259
1238872273
1239418524
1240283215
1240870047
1241789227
1242477751
1243039354
1243670182
1244554085

e. Why or why not?

More than 1% of the data was changed when adding five observations.
7. Generating a Systematic Sample
Generate a systematic sample by selecting every tenth supplier starting with observation 10 from the
data set orion.product_dim.
a. The sample should contain only the variables Product_Line, Product_ID, Product_Name, and
Supplier_Name.
b. Name the output data set products_sample.
c. Print the first five observations of products_sample. Omit the observation numbers.
3-78 Chapter 3 Accessing Observations

p303s07
data products_sample;
do i=10 to TotObs by 10;
set orion.product_dim(keep=Product_Line Product_ID
Product_Name Supplier_Name)
nobs=TotObs
point=i;
output;
end;
stop;
run;

proc print data=products_sample(obs=5) noobs;

title "Systematic Sample of Products";
run;
8. Generating a Random Sample with Replacement
Generate a random sample with replacement of 50 customers from orion.customer_dim. If the
customer age is less than 40, place those customers in a data set named underforty. If the customer
age is greater than or equal to 40, place those customers in a data set named fortyplus.
p303s08
data underforty fortyplus;
drop i SampSize;
SampSize=50;
do i=1 to SampSize;
PickIt=ceil(ranuni(0) * TotObs);
set orion.customer_dim point=PickIt nobs=TotObs;
if Customer_Age < 40 then output underforty;
else output fortyplus;
end;
stop;
run;

proc print data=fortyplus;

title 'Customers Forty or Older';
run;

proc print data=underforty;

title 'Customers under Forty';
run;
 3.5 Solutions 3-79

9. Generating a Random Sample without Replacement

Generate a random sample without replacement of approximately 10% of the data in
orion.customer_dim.
p303s09
data sample(drop=ObsLeft SampSize);
SampSize=int(.10 * TotObs);
ObsLeft=TotObs;
do while(SampSize > 0 and ObsLeft > 0);
PickIt + 1;
if ranuni(0) < SampSize / ObsLeft then
do;
ObsPicked=PickIt;
set orion.customer_dim point=PickIt
nobs=TotObs;
output;
SampSize=SampSize - 1;
end;
ObsLeft=ObsLeft - 1;
end;
stop;
run;

/* to get an approximate 10% sample */

data sample;
set orion.customer_dim;
if ranuni(0) <= .10;
run;
3-80 Chapter 3 Accessing Observations

Solutions to Student Activities (Polls/Quizzes)

3.02 Multiple Choice Poll – Correct Answer

If a WHERE statement uses an index to retrieve a small
subset of data, which of these resources is conserved?
a. I/O
b. Disk space
c. Memory
d. Programmer time

34

3.03 Multiple Answer Poll – Correct Answers

On which of these indexed variables can you assign the
UNIQUE option?
a. Customer_ID in an orders data set where
a customer can place multiple orders
b. Order_Date in an orders data set
c. Employee_ID in a data set containing each
individual employee and the family members’
names stored in variables Dependent1 –
Dependent10
d. Product_ID in a data set containing the
product identifier and the product description

43
 3.5 Solutions 3-81

3.04 Quiz – Correct Answer

Open and submit the program p303a01.
What error messages are in the log?
1 options msglevel=n;
2 proc datasets library=orion nolist;
3 modify sales_history;
4 index create Customer_ID;
ERROR: An index named Customer_ID with the same definition already exists for
file ORION.SALES_HISTORY.DATA.
5 index create Product_Group;
6 index create SaleID=(Order_ID
7 Product_ID)/unique;
8 quit;

NOTE: Statements not processed because of errors noted above.

NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE DATASETS used (Total process time):
real time 0.48 seconds
cpu time 0.09 seconds

If an index exists, it must be deleted before it can be

re-created.
54
3-82 Chapter 3 Accessing Observations

3.05 Multiple Answer Poll – Correct Answers

Which of the following WHERE conditions could possibly
use an index?
a. where Product_ID=220100300042;
b. where Customer_ID ne 3245;
c. where Customer_ID=15020 or
Customer_ID=14853;
d. where Order_ID=1230036183;
e. where Customer_ID='3245';

72

a. No, there is no index on Product_ID.

b. Yes, although this would be a large subset, an index might be used.
c. Yes, SAS converts the WHERE statement to the following code:

where Customer_ID in (15020, 14853);

d. Yes, Order_ID is the primary key variable in the SaleID index, so that index could be used.
e. This statement would not execute because there is a syntax error. The WHERE statement requires a
numeric constant (3245) because Customer_ID is a numeric variable.
 3.5 Solutions 3-83

3.06 Multiple Choice Poll – Correct Answer

When does the subsetting IF statement select
observations?
a. before the observation is copied into the PDV
b. after the observation is in the PDV

76

3.07 Multiple Choice Poll – Correct Answer

Which of the following WHERE statements can use the
composite index SaleID for compound optimization?
a. where Order_ID=240200100038 or
Product_ID=1230151326;
b. where Order_ID=. and
Product_ID=1230151326;
c. where int(Order_ID/1000000000)=240
and Product_ID=1230151326;
d. where Order_ID>240000000000 and
Product_ID<1240000000;

83

a. No, the WHERE statement uses the OR comparison operator.

b. Yes, the WHERE statement can use SALEID. The conditions for compound optimization are met.
c. No, the WHERE statement uses a function on the variable Order_ID that is not the TRIM or
SUBSTR function.
d. No, the WHERE statement does not use either the EQ or IN operator in one of the conditions.
3-84 Chapter 3 Accessing Observations

3.08 Multiple Choice Poll – Correct Answer

Which of the following is used to determine the I/O to read
a SAS data set sequentially?
a. the page size of the input data set and the number
of buffers available
b. the number of observations and the number
of variables
c. the page size of the output data set and the number
of output buffers available

91

3.09 Quiz – Correct Answer

Are POINT= and NOBS= individual statements
or part of the SET statement?
data subset;
do PickIt=1 to TotObs by 50;
set orion.order_fact(keep=Customer_ID
Employee_ID Street_ID Order_ID)
point=PickIt
nobs=TotObs;
output;
end;
stop;
run;

POINT= and NOBS= are part of the SET statement.

p303d07
113
 3.5 Solutions 3-85

3.10 Poll – Correct Answer

Instead of the CEIL function, would the INT function return
the same results?
€ Yes
€ No

The INT function returns the integer portion of its

argument, which could possibly be 0 and never be 5.

136
3-86 Chapter 3 Accessing Observations

Solutions to Chapter Review

Chapter Review – Correct Answers

1. What is one purpose of an index?
An index can be used to perform any of these
tasks:
„ yield faster access to small subsets (WHERE)

„ return observations in sorted order (BY)

„ perform table lookup operations (SET with

KEY=)
„ join observations (PROC SQL)

„ modify observations (MODIFY with KEY=)

2. What are the three ways to create an index?

The INDEX= data set option, the DATASETS
procedure, and the SQL procedure
151 continued...

Chapter Review – Correct Answers

3. What SAS system option is used to view information
about index usage in the log?
MSGLEVEL=I
4. How can you tell whether a SAS data set has an
index?
Use the CONTENTS or DATASETS procedure, the
PROPERTIES window from the SAS Explorer, or
SAS Management Console.
5. Which functions can use an index?
TRIM and SUBSTR (under the condition that the
second argument must be 1)
6. Does a subsetting IF use an index?
No
152 continued...
 3.5 Solutions 3-87

Chapter Review – Correct Answers

7. Which DATA set option forces SAS to use an index for
WHERE clause processing?
IDXWHERE=YES or IDXNAME=index-name
8. Does sorting a data set before indexing help the index
perform better?
Yes
9. Does a DATA step using a SET statement that reads
and writes the same data set automatically maintain
an index?
No

153
3-88 Chapter 3 Accessing Observations
Chapter 4 Introduction to Lookup
Techniques

4.1 Introduction to Lookup Techniques.............................................................................. 4-3

4.2 In-Memory Lookup Techniques ..................................................................................... 4-5

4.3 Disk Storage Techniques ............................................................................................. 4-13

4.4 Chapter Review............................................................................................................. 4-28

4.5 Solutions ....................................................................................................................... 4-29

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 4-29

Solutions to Chapter Review ................................................................................................ 4-32

4-2 Chapter 4 Introduction to Lookup Techniques
 4.1 Introduction to Lookup Techniques 4-3

4.1 Introduction to Lookup Techniques

Objectives
„ Define table lookup.
„ List table lookup techniques.

Table Lookups
Lookup values for a table lookup can be stored in the
following: Lookup Values
„ array

„ hash object

„ format

„ data set

Lookup techniques include

the following:
Data Values
„ array subscript value

„ hash object key value

„ FORMAT statement,
PUT function
„ MERGE, SET/SET, join
4

 The hash object is new in SAS®9.

4-4 Chapter 4 Introduction to Lookup Techniques

4.01 Multiple Choice Poll

Which of these is an example of a table lookup?
a. You have the data for January sales in one data set,
February sales in a second data set, and March sales
in a third. You need to create a report for the entire
first quarter.
b. You want to send birthday cards to employees.
The employees’ names and addresses are in one
data set and their birthdates are in another.
c. You need to calculate the amount each customer
owes for his purchases. The price per item and the
number of items purchased are stored in the same
data set.

Overview of Table Lookup Techniques

„ Arrays, hash objects, and formats provide an
in-memory lookup table.
„ The DATA step MERGE statement, multiple SET
statements in the DATA step, and SQL procedure
joins use lookup values that are stored on disk.

8
 4.2 In-Memory Lookup Techniques 4-5

4.2 In-Memory Lookup Techniques

Objectives
„ Describe arrays as a lookup technique.
„ Describe hash objects as a lookup technique.
„ Describe formats as a lookup technique.

10

4.02 Multiple Answer Poll

Which techniques do you currently use when you perform
table lookups with a single data set?
a. Arrays
b. Hash object
c. Formats
d. None of the above

12
4-6 Chapter 4 Introduction to Lookup Techniques

Overview of Arrays
An array is similar to a numbered row of buckets.

1 2 3 4

„ SAS puts a value in a bucket based on the bucket

number.
„ A value is retrieved from a bucket based on the bucket
number.

15

Overview of Arrays
General form of the ARRAY statement:

DATA data-set-name;
ARRAY array-name { subscript } <$><length>
<array-elements> <(initial-value-list)>;
< READ statement (s)>
new-variable=array-name{subscript-value};
RUN;
The ARRAY statement
associates variables or The assignment statement
initial values to be retrieved retrieves values from the
using the array name and a array based on the value of
subscript value. the subscript.

 The READ statement can be the SET, MERGE or

16
INFILE/INPUT statement.
 4.2 In-Memory Lookup Techniques 4-7

Overview of Arrays
data country_info;
array Cont_Name{91:96} $ 30 _temporary_
('North America',
' ',
'Europe',
'Africa',
'Asia',
'Australia/Pacific');
set orion.country;
Continent=Cont_Name{Continent_ID};
run;

The ARRAY statement The assignment

associates variables or statement retrieves
initial values to be retrieved values from the array
using the array name and a based on the value of
subscript value. the subscript.
17 p304d01

Setup for the Poll

p304d01
data country_info;
array Cont_Name{91:96} $ 30 _temporary_
('North America',
' ',
'Europe',
'Africa',
'Asia',
'Australia/Pacific');
set orion.country;
Continent=Cont_Name{Continent_ID};
run;

19
4-8 Chapter 4 Introduction to Lookup Techniques

4.03 Multiple Choice Poll

In p304d01, how many elements are in the array
Cont_name?
a. 0
b. 5
c. 6
d. unknown

20

Overview of a Hash Object

A hash object is similar to rows of buckets that are
identified by the value of a key.
Key Data Data
„ SAS puts value(s)
in the data
bucket(s) based on
the value(s) in the
key bucket.
„ Value(s) are
retrieved from the
data bucket(s)
based on the
value(s) in the key
bucket.
24
 4.2 In-Memory Lookup Techniques 4-9

Overview of Hash Objects

General form of the hash object:
The syntax within the DO
DATA data-set-name; group defines and can
< READ statement(s) > populate the hash object.
IF _N_=1 THEN DO;
DECLARE HASH object-name(<attribute:value>);
object-name.DEFINEKEY('key-name');
object-name.DEFINEDATA('data-name');
The FIND method
object-name.DEFINEDONE(); retrieves the data
END; value based on
the key value.
return-code=object-name.FIND(<key: value>);
RUN;

 The READ statement can be the SET, MERGE,

25
or INFILE/INPUT statement.

Overview of Hash Objects

The syntax within
the DO group defines
data country_info; and populates the
length Continent_Name $ 30; hash object.
if _N_=1 then do;
declare hash Cont_Name(dataset:'orion.continent');
Cont_Name.definekey('Continent_ID');
Cont_Name.definedata('Continent_Name');
Cont_Name.definedone();
end;
set orion.country;
rc=Cont_Name.find(key:Continent_ID);
if rc=0;
run; The FIND method
retrieves the data
value based on
the key value.

p304d02
26
4-10 Chapter 4 Introduction to Lookup Techniques

Setup for the Poll

p304d02
data country_info;
length Continent_Name $ 30;
if _N_=1 then do;
declare hash Cont_Name(dataset:'orion.continent');
Cont_Name.definekey('Continent_ID');
Cont_Name.definedata('Continent_Name');
Cont_Name.definedone();
end;
set orion.country;
rc=Cont_Name.find(key:Continent_ID);
if rc=0;
run;

28

4.04 Multiple Choice Poll

In p304d02, how many times do the statements
in the DO group execute?
a. only once
b. once for every observation in the data set
orion.country
c. once for every observation in the data set
orion.continent

29
 4.2 In-Memory Lookup Techniques 4-11

Overview of a Format
A format is similar to rows of buckets that are identified
by the data value.
Data Value Label „ SAS puts data values and
label values in the buckets
when the format is used in a
FORMAT statement, PUT
function, or PUT statement.
„ SAS uses a binary search
on the data value bucket in
order to return the value in
the label bucket.

33

Overview of a Format
General form of the user-defined format:
The FORMAT step
PROC FORMAT; compiles the format
VALUE <$>fmtname range-1=label-1 and stores it on disk.
...
range-n=label-n;
RUN;
When the PUT
DATA data-set-name; function executes,
< READ statement(s)>; the format is loaded
new-variable=PUT(variable,fmtname.); into memory, and a
RUN; binary search is
used to retrieve the
format value.

 The READ statement can be the SET, MERGE, or

34
INFILE/INPUT statement.
4-12 Chapter 4 Introduction to Lookup Techniques

Overview of a Format
The FORMAT step
proc format; compiles the format
value Cont_Name and stores it on disk.
91='North America'
93='Europe'
94='Africa'
95='Asia'
96='Australia/Pacific';
run;

data country_info;
set orion.country;
Continent=put(Continent_ID,Cont_Name.);
run;

When the PUT function executes, the format

is loaded into memory, and a binary search
is used to retrieve the format value. p304d03
35
 4.3 Disk Storage Techniques 4-13

4.3 Disk Storage Techniques

Objectives
„ List methods for combining data horizontally.
„ Use multiple SET statements to combine data
horizontally.
„ Compare methods for combining SAS data sets.

37

Combining Data Horizontally

DATA step techniques for combining data horizontally
include using the following:
„ MERGE statement

„ multiple SET statements

„ UPDATE statement

„ MODIFY statement

In addition, you can use the SQL procedure with an inner

or outer join.

38
4-14 Chapter 4 Introduction to Lookup Techniques

4.05 Multiple Answer Poll

Which techniques do you currently use when you perform
table lookups with multiple data sets?
a. MERGE statement
b. Joins
c. Multiple SET statements
d. UPDATE statement
e. MODIFY statement
f. None of the above

40

Overview of Merges and Joins

The DATA step MERGE and the SQL join operators are
similar to multiple stacks of buckets that are referred to
by the value of one or more common variables.
By Value(s) Data Data By Value(s) Data Data

41
 4.3 Disk Storage Techniques 4-15

DATA Step MERGE Statement

General form of the DATA step merge:
DATA data-set-name;
MERGE SAS-data-sets;
BY variables;
RUN;

Matches on equal values for like-named variables:

Continent_ID Continent_ID

Continent_ID

42

DATA Step MERGE Statement

proc sort data=orion.country out=country;
by Continent_ID;
run;

data country_info;
merge country orion.continent;
by Continent_ID;
run;
Matches on equal
values for like-named
variables

p304d04
43
4-16 Chapter 4 Introduction to Lookup Techniques

Setup for the Poll

p304d04
proc sort data=orion.country out=country;
by Continent_ID;
run;

data country_info;
merge country orion.continent;
by Continent_ID;
run;

45

4.06 Multiple Choice Poll

In p304d04, if the data set country has seven
observations and the data set orion.continent has five
observations, what stops the execution of the DATA step?
a. end of file for work.country, the data set with the
most observations
b. end of file for orion.continent, the last data set listed
in the MERGE statement
c. end of file for the data set that contains the final value
of the BY variable Continent_ID

46
 4.3 Disk Storage Techniques 4-17

The SQL Procedure

You can use an SQL procedure inner or outer join
to create a SAS data set.
General form of the SQL procedure CREATE TABLE
statement with an inner join:

PROC SQL;
CREATE TABLE SAS-data-set AS
SELECT column-1, column-2,… ,column-n
FROM table-1, table-2,…,table-n
WHERE joining criteria
ORDER BY sorting criteria;
QUIT;
Performs an inner join based
on the WHERE criteria

48

The SQL Procedure

proc sql;
create table country_info as
select country.*, Continent_Name
from orion.country, orion.continent
where country.Continent_ID=
continent.Continent_ID;
order by country.Continent_ID;
quit;

Performs an inner join where

the Continent_ID values from
both data sets are equal

p304d05
49
4-18 Chapter 4 Introduction to Lookup Techniques

4.07 Multiple Choice Poll

Which of the following is true of the SQL inner join?
a. The resulting data set contains only the observations
with matching key values.
b. The resulting data set contains both the observations
with matching key values and those observations
where the key values do not match.

51

Multiple SET Statements

The DATA step with multiple SET statements combines
data sets by performing one-to-one reading.
Data Data Data Data

53
 4.3 Disk Storage Techniques 4-19

Multiple SET Statements

You can use multiple SET statements to combine
observations from several SAS data sets.
When you use multiple SET statements, the following
occurs:
„ Processing stops when SAS encounters the end-of-file
marker on either data set.
„ The variables in the PDV are not reinitialized when
a second SET statement is executed.

54

Multiple SET Statements

General form of the DATA step with multiple set
statements:

DATA data-set-name;
SET SAS-data-set;
SET SAS-data-set;
RUN;

55
4-20 Chapter 4 Introduction to Lookup Techniques

Multiple SET Statements

data country_info;
set orion.country;
set orion.continent;
run;

Listing of country_info
Country_ Country_ Continent_ Country_Former
Obs Country Name Population ID ID Name Continent_Name

1 AU Australia 20,000,000 160 91 North America

2 CA Canada . 260 93 Europe
3 DE Germany 80,000,000 394 94 East/West Germany Africa
4 IL Israel 5,000,000 475 95 Asia
5 TR Turkey 70,000,000 905 96 Australia/Pacific

p304d06
56

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total _N_
1 2 . 1

57 ...
 4.3 Disk Storage Techniques 4-21

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total D _N_
1 2 A . 1

58 ...

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total D _N_
1 2 A 3 1

59 ...
4-22 Chapter 4 Introduction to Lookup Techniques

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;
Implicit OUTPUT;
Implicit RETURN;
PDV
X Y Z Total D _N_
1 2 A 3 1

60 ...

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;
Initialize PDV.

PDV
X Y Z Total D _N_
1 2 A . 2

61 ...
 4.3 Disk Storage Techniques 4-23

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total D _N_
2 3 A . 2

62 ...

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total D _N_
2 3 B . 2

63 ...
4-24 Chapter 4 Introduction to Lookup Techniques

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total D _N_
2 3 B 5 2

64 ...

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;
Implicit OUTPUT;
Implicit RETURN;
PDV
X Y Z Total D _N_
2 3 B 5 2

65 ...
 4.3 Disk Storage Techniques 4-25

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;
Initialize PDV.

PDV
X Y Z Total D _N_
2 3 B . 3

66 ...

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
run;

PDV
X Y Z Total D _N_
3 4 B . 3

67 ...
4-26 Chapter 4 Introduction to Lookup Techniques

Execution
one two
X Y Z data three;
1 2 A set one;
2 3 B set two;
3 4 Total=X+Y;
EOF run;

Processing stops.
PDV
X Y Z Total D _N_
3 4 B . 3

three
X Y Z Total
1 2 A 3
2 3 B 5
68

Setup for the Poll

The previous example created a data set named three
with two observations.
one two data three;
X Y Z set one;
1 2
set two;
A
Total=X+Y;
2 3 B run;
3 4
Using the same one and two data sets, if the SET
statements were reversed, how many observations would
be in the data set three?
data three;
set two;
set one;
Total=X+Y;
run;
70
 4.3 Disk Storage Techniques 4-27

4.08 Multiple Choice Poll

Using the same one and two data sets, if the SET
statements were reversed, how many observations would
be in the data set three?
a. 5
b. 2
c. 3
d. 6

71

DATA Step Methods for Reading SAS Data

Code Which variables are reinitialized What stops the
to missing at the top DATA step?
of the DATA step?
data two; variables created in the DATA step end of the file
set one;
New_Var=Value; for data set one
run;
data three; „ variables created in the DATA the last end of
merge one two;
by Var; step file that is
New_Var=Value; „ all variables when the BY value encountered
run;
changes
data three; „ variables created in the DATA end of the file
set one two;
New_Var=Value; step for data set two
run; „ all variables when SAS finishes
reading data set one and starts
reading data set two
data three; variables created in the DATA step the first end of
set one;
set two; file that is
New_Var=Value; encountered
73 run;
4-28 Chapter 4 Introduction to Lookup Techniques

4.4 Chapter Review

Chapter Review
1. What are the three types of in-memory table lookups?

2. What are three types of disk storage table lookups?

3. When multiple SET statements are executed, when

does execution stop?

74
 4.5 Solutions 4-29

4.5 Solutions

Solutions to Student Activities (Polls/Quizzes)

4.01 Multiple Choice Poll – Correct Answer

Which of these is an example of a table lookup?
a. You have the data for January sales in one data set,
February sales in a second data set, and March sales
in a third. You need to create a report for the entire
first quarter.
b. You want to send birthday cards to employees.
The employees’ names and addresses are in one
data set and their birthdates are in another.
c. You need to calculate the amount each customer
owes for his purchases. The price per item and the
number of items purchased are stored in the same
data set.

4.03 Multiple Choice Poll – Correct Answer

In p304d01, how many elements are in the array
Cont_name?
a. 0
b. 5
c. 6
d. unknown

21
4-30 Chapter 4 Introduction to Lookup Techniques

4.04 Multiple Choice Poll – Correct Answer

In p304d02, how many times do the statements
in the DO group execute?
a. only once
b. once for every observation in the data set
orion.country
c. once for every observation in the data set
orion.continent

30

4.06 Multiple Choice Poll – Correct Answer

In p304d04, if the data set country has seven
observations and the data set orion.continent has five
observations, what stops the execution of the DATA step?
a. end of file for work.country, the data set with the
most observations
b. end of file for orion.continent, the last data set listed
in the MERGE statement
c. end of file for the data set that contains the final value
of the BY variable Continent_ID

47
 4.5 Solutions 4-31

4.07 Multiple Choice Poll – Correct Answer

Which of the following is true of the SQL inner join?
a. The resulting data set contains only the observations
with matching key values.
b. The resulting data set contains both the observations
with matching key values and those observations
where the key values do not match.

52

4.08 Multiple Choice Poll – Correct Answer

Using the same one and two data sets, if the SET
statements were reversed, how many observations would
be in the data set three?
a. 5
b. 2
c. 3
d. 6

72
4-32 Chapter 4 Introduction to Lookup Techniques

Solutions to Chapter Review

Chapter Review – Correct Answers

1. What are the three types of in-memory table lookups?
arrays, hash objects, and formats
2. What are three types of disk storage table lookups?
PROC SQL, the DATA step with a MERGE
statement, or the DATA step with multiple SET
statements
3. When multiple SET statements are executed, when
does execution stop?
Execution stops when the first end of file is
encountered.

75
Chapter 5 Using DATA Step Arrays

5.1 Using One-Dimensional Arrays ..................................................................................... 5-3

Exercises .............................................................................................................................. 5-18

5.2 Using Multidimensional Arrays ................................................................................... 5-22

Exercises .............................................................................................................................. 5-35

5.3 Loading a Multidimensional Array from a SAS Data Set .......................................... 5-40
Exercises .............................................................................................................................. 5-63

5.4 Chapter Review............................................................................................................. 5-69

5.5 Solutions ....................................................................................................................... 5-70

Solutions to Exercises .......................................................................................................... 5-70

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 5-79

Solutions to Chapter Review ................................................................................................ 5-83

5-2 Chapter 5 Using DATA Step Arrays
 5.1 Using One-Dimensional Arrays 5-3

5.1 Using One-Dimensional Arrays

Objectives
„ Define one-dimensional arrays.
„ Use a one-dimensional array for a table lookup task.

Overview of Arrays (Review)

An array is similar to a row of numbered buckets.

1 2 3 4

„ SAS puts a value in a bucket based on the bucket

number.
„ A value is retrieved from a bucket based on the bucket
number.

4
5-4 Chapter 5 Using DATA Step Arrays

Defining Arrays (Review)

An array is a temporary grouping of SAS variables that
are arranged in a particular order and identified by an
array name.
The following tasks can be accomplished using an array:
„ performing repetitive calculations on a group
of variables
„ creating many variables with the same attributes

„ restructuring data

„ performing a table lookup with one or more numeric

factors

 An array exists only for the duration of the current

DATA step.
5
 5.1 Using One-Dimensional Arrays 5-5

Using One-Dimensional Arrays (Review)

To use an array, declare the array by using an ARRAY
statement.
General form of the one-dimensional ARRAY statement:

ARRAY array-name {number-of-elements} <$> <length>

<list-of-variables> <(initial-values)>;

is a SAS name that identifies the group of variables.

array-name

number-of- elements is the number of variables in the group. You must enclose this value in
parentheses, braces, or brackets.

$ indicates that the elements in the array are character elements.

length specifies the length of elements in the array that were not previously
assigned a length.

list-of-variables is a list of the names of the variables in the group. All variables that are
defined in a given array must be of the same type, either all character or
all numeric.

initial-values gives initial values for the corresponding positional elements in the array.

_TEMPORARY_ is a keyword that can be used instead of list-of-variables to define an

array that is not associated with program data vector variables. Arrays of
temporary elements are useful when the only purpose for creating
an array is to perform a calculation. To preserve the result of the
calculation, assign it to a variable. You can improve performance
time by using temporary data elements.
5-6 Chapter 5 Using DATA Step Arrays

Using One-Dimensional Arrays (Review)

Examples of an ARRAY statement follow.
array numarray{3} Num1 – Num3;

array char{4} $ 6; Array Name

array num{5} _temporary_ (5, 6, 7, 8, 9);

array yr{2000:2002} Yr2000 Yr2001 Yr2002;

8 ...

Using One-Dimensional Arrays (Review)

Examples of an ARRAY statement follow.
array numarray{3} Num1 – Num3;

array char{4} $ 6; number of elements

array num{5} _temporary_ (5, 6, 7, 8, 9);

array yr{2000:2002} Yr2000 Yr2001 Yr2002;

9 ...
 5.1 Using One-Dimensional Arrays 5-7

Using One-Dimensional Arrays (Review)

Examples of an ARRAY statement follow.
array numarray{3} Num1 – Num3;
names three numeric variables

array char{4} $ 6;
creates four character variables, char1 – char4, each a length of 6

array num{5} _temporary_ (5, 6, 7, 8, 9);

creates temporary numeric elements, and stores the numeric values 5, 6, 7, 8, 9

array yr{2000:2002} Yr2000 Yr2001 Yr2002;

names three numeric variables

10

5.01 Multiple Choice Poll

How many elements are referenced by the following
ARRAY statement?

array numarray{*} Num1 – Num12;

a. 0
b. 1
c. 12
d. Unknown

12
5-8 Chapter 5 Using DATA Step Arrays

The DIM Function

You can use the DIM function to return the number of
elements in an array. For example, use the DIM function
to provide the end value for a DO loop.
array numarray{*} Num1 – Num12;
<additional statements>
do i=1 to dim(numarray);
<additional statements>
end;

Equivalent code:
array numarray{12} Num1 – Num12;
<additional statements>
do i=1 to 12;
<additional statements>
end;
14

The DIM function returns the number of elements in a one-dimensional array.

General form of the DIM function:

DIM(array-name)

array-name specifies the name of an array that was previously

defined in the same DATA step.

Business Scenario
The data set orion.employee_payroll contains each
employee’s hired date and current salary.

Partial Listing of orion.employee_payroll

Employee Employee Employee_ Employee_ Marital_
Salary Birth_Date Dependents
_ID _Gender Hire_Date Term_Date Status

120101 M 163040 18AUG1976 01JUL2003 . S 0

120102 M 108255 11AUG1969 01JUN1989 . O 2
120103 M 87975 22JAN1949 01JAN1974 . M 1
120104 F 46230 11MAY1954 01JAN1981 . M 1
120105 F 27110 21DEC1974 01MAY1999 . S 0
. . . . . . . .
. . . . . . . .
. . . . . . . .

15
 5.1 Using One-Dimensional Arrays 5-9

Business Scenario
The data set orion.salary_stats contains statistics for all
Orion Star employees for the years 1974 through 2007.
For example, the average salary of the employees hired
in 1974 is currently $39,243.61.
Partial Listing of orion.salary_stats
Statistic Yr1974 Yr1975 Yr1976 . . . Yr2006 Yr2007
Num_of_Emps 61 4 6 . . . 97 3
Median_Salary 30025 29442.5 30020 . . . 26970 27240
Std_Salary 28551.9 9918.35 22356.91 . . . 2579.67 2922.12
Sum_Salary 2393860 132150 235030 . . . 2704720 86585
Avg_Salary 39243.61 33037.5 39171.67 . . . 27883.71 28861.67

16

Business Scenario
The two data sets must be combined to calculate the
difference between the average salary and the actual
current salary for each employee based on the year
of hire.
Partial Listing of compare
Using One Dimensional Arrays

Year_
Obs Employee_ID Hired Salary Average Salary_Dif
1 120101 2003 $163,040.00 $35,082.50 $127,957.50
2 120102 1989 $108,255.00 $88,588.75 $19,666.25
3 120103 1974 $87,975.00 $39,243.61 $48,731.39
4 120104 1981 $46,230.00 $36,436.67 $9,793.33
5 120105 1999 $27,110.00 $36,533.75 $-9,423.75
6 120106 1974 $26,960.00 $39,243.61 $-12,283.61
7 120107 1974 $30,475.00 $39,243.61 $-8,768.61
8 120108 2006 $27,660.00 $27,883.71 $-223.71

17
5-10 Chapter 5 Using DATA Step Arrays

Setup for the Poll

The two data sets that need to be combined are as
follows:
Partial Listing of orion.salary_stats
Statistic Yr1974 Yr1975 Yr1976 . . . Yr2006 Yr2007
Avg_Salary 39243.61 33037.5 39171.67 . . . 27883.71 28861.67

Partial Listing of orion.employee_payroll

Employee Employee Birth_ Employee_
Salary . . .
_ID _Gender Date Hire_Date

120101 M 163040 18AUG1976 01JUL2003 . . .

120102 M 108255 11AUG1969 01JUN1989 . . .
120103 M 87975 22JAN1949 01JAN1974 . . .
120104 F 46230 11MAY1954 01JAN1981 . . .
120105 F 27110 21DEC1974 01MAY1999 . . .
19

5.02 Poll
Can the two data sets be merged with the DATA step
MERGE statement or joined with the SQL procedure
without pre-processing the data?
€ Yes
€ No

20
 5.1 Using One-Dimensional Arrays 5-11

5.03 Poll
What do the two data sets have in common?
€ They have the year in common.
€ They have nothing in common.

22

Using a One-Dimensional Array

data compare;
keep Employee_ID Year_Hired Salary Average
Salary_Dif;
format Salary Average Salary_Dif dollar12.2;
c array yr{1974:2007} Yr1974-Yr2007;
d if _N_=1 then set orion.salary_stats
(where=(Statistic='Avg_Salary'));
set orion.employee_payroll
(keep=Employee_ID
Employee_Hire_Date
Salary);
Year_Hired=year(Employee_Hire_Date);
e Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

p305d01
24

c The array yr is associated with the variables Yr1974, Yr1975, Yr1976, and so forth through
YR2007.

d Read only the observation where the value of the variable Statistic is Avg_Salary.
e The value of the element on the yr array is referenced positionally by the value of the variable
Year_Hired and is assigned to the variable Average.
5-12 Chapter 5 Using DATA Step Arrays

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
. . . . . . . .

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
. . . . . . . 1
25 ...

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
. . . 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
.
35082.5 .
29904.44 .
28861.67 .
Avg_Salary . . . 1
26 ...
 5.1 Using One-Dimensional Arrays 5-13

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 . . 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 . 1
27 ...

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 . . 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 2003 1
28 ...
5-14 Chapter 5 Using DATA Step Arrays

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 . . 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 2003 1
29 ...

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{2003};
Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 35082.5 . 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 2003 1
30 ...
 5.1 Using One-Dimensional Arrays 5-15

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 35082.5 127957.5 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 2003 1
31 ...

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
Implicit OUTPUT;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 Implicit RETURN;
if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 35082.5 127957.5 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 2003 1
32 ...
5-16 Chapter 5 Using DATA Step Arrays

Execution Partial Listing of orion.salary_stats

Statistic Yr1974 Yr1975 Yr1976 . . .
Partial Listing of
Avg_Salary 39243.61 33037.5 39171.67 . . .
orion.employee_payroll
Employee Employee_ data compare;
Salary keep Employee_ID Year_Hired Salary Average
_ID Hire_Date Salary_Dif;
120101 01JUL2003 163040 format Salary Average Salary_Dif dollar12.2;
array yr{1974:2007} Yr1974-Yr2007;
120102 01JUN1989 108255 if _N_=1 then set orion.salary_stats
120103 01JAN1974 87975 (where=(Statistic='Avg_Salary'));
set orion.employee_payroll
120104 01JAN1981 46230 (keep=Employee_ID Employee_Hire_Date Salary);
Year_Hired=year(Employee_Hire_Date);
120105 01MAY1999 27110 Average=yr{Year_Hired};
Salary_Dif=Salary-Average; Continue until EOF.
run;

Partial PDV yr{1974} yr{1975} yr{1976} yr{1977} yr{1978}

Salary_
Salary Average D Yr1974 DYr1975 D Yr1976 DYr1977 DYr1978 . . .
Dif
163040 35082.5 127957.5 39243.61 33037.5 39171.67 34170 37506.25

yr{2003} yr{2004} yr{2007}

Employee Employee_ Year_
D Yr2003 DYr2004 . . . DYr2007 DStatistic D D _N_
_ID Hire_Date Hired
35082.5 29904.44 28861.67 Avg_Salary 120101 01JUL2003 2003 1
33

Resulting Data
proc print data=compare(obs=8);
var Employee_ID Year_Hired Salary Average Salary_Dif;
title 'Using One Dimensional Arrays';
run;

PROC PRINT Output

Using One Dimensional Arrays

Year_
Obs Employee_ID Hired Salary Average Salary_Dif
1 120101 2003 $163,040.00 $35,082.50 $127,957.50
2 120102 1989 $108,255.00 $88,588.75 $19,666.25
3 120103 1974 $87,975.00 $39,243.61 $48,731.39
4 120104 1981 $46,230.00 $36,436.67 $9,793.33
5 120105 1999 $27,110.00 $36,533.75 $-9,423.75
6 120106 1974 $26,960.00 $39,243.61 $-12,283.61
7 120107 1974 $30,475.00 $39,243.61 $-8,768.61
8 120108 2006 $27,660.00 $27,883.71 $-223.71

p305d01
34
 5.1 Using One-Dimensional Arrays 5-17

5.04 Multiple Answer Poll

Which of the following ARRAY statements are similar
to the statement
array yr{1974:2007} Yr1974-Yr2007;

and will compile without errors?

a. array yr{34} Yr1974-Yr2007;
b. array yr{1974-2007} Yr1974-Yr2007;
c. array yr{74:07} Yr1974-Yr2007;
d. array yr{74-07} Yr1974-Yr2007;
e. array yr{*} Yr1974-Yr2007;

36

Using either of the alternative ARRAY statements, you must change the array reference that creates the
variable Average.
p305d01a
data compare;
keep Employee_ID Year_Hired Salary Average Salary_Dif;
format Salary Average Salary_Dif dollar12.2;
c array yr{34} Yr1974-Yr2007;
if _N_=1 then
set orion.salary_stats(where=(Statistic='Avg_Salary'));
set orion.employee_payroll;
d Year_Hired=year(Employee_Hire_Date)-1973;
Average=yr{Year_Hired};
Salary_Dif=Salary-Average;
run;
c The array yr is associated with the variables Yr1974, Yr1975, Yr1976, and so forth through
YR2007.

d Because the subscript values for the yr array are 1 to 34, adjust the Year_Hired variable so that
yr{1} corresponds to 1974, yr{2} corresponds to 1975, and so forth. The value of the element of the
yr array is referenced positionally by the value of the variable Year_Hired and is assigned to the
variable Average.
5-18 Chapter 5 Using DATA Step Arrays

Exercises

Level 1

1. Using a One-Dimensional Array to Combine Data

The data set orion.retail has information about retail sales.
Partial Listing of orion.retail
Partial orion.retail Data Set

Order_ Delivery_
Obs Customer_ID Employee_ID Street_ID Date Date Order_ID

1 63 121039 9260125492 11JAN2003 11JAN2003 1230058123

2 41 120174 1600101527 28JAN2003 28JAN2003 1230147441
3 183 120134 1600100760 27FEB2003 27FEB2003 1230315085
4 56 121059 9260111871 15MAR2003 15MAR2003 1230404278
5 183 120149 1600100760 22MAR2003 22MAR2003 1230440481

Total_Retail_ CostPrice_
Obs Product_ID Quantity Price Per_Unit Discount

1 220101300017 1 $16.50 $7.45 .

2 240600100010 2 $32.00 $6.50 .
3 240200200039 3 $63.60 $8.80 .
4 220200300002 2 $75.00 $17.05 .
5 230100600005 1 $129.80 $63.20 .

The data set orion.retail_information has statistics about those retail sales.
Partial Listing of orion.retail_information
Partial orion.retail_information Data Set

Obs Statistic Month1 Month2 Month3

1 Sum_Retail_Price $1,599.80 $1,160.80 $113.70

2 Mean_Retail_Price $228.54 $193.47 $28.43
3 Median_Retail_Price $258.20 $123.30 $27.65

Obs Month4 Month5 Month6 Month7 Month8

1 $671.10 $520.70 $561.99 $288.29 $1,033.40

2 $83.89 $86.78 $70.25 $48.05 $103.34
3 $49.10 $68.30 $52.20 $44.90 $54.25

Obs Month9 Month10 Month11 Month12

1 $425.70 $736.10 $2,399.30 $1,347.58

2 $70.95 $105.16 $218.12 $134.76
3 $65.35 $101.50 $69.40 $61.70
 5.1 Using One-Dimensional Arrays 5-19

a. Combine the two data sets to create a data set named compare. The data set should contain the
variables from orion.retail and variables named Month and Median_Retail_Price, where
Month is the month of the date that the product was ordered.
b. Print the first eight observations of the resulting data set.
PROC PRINT Output
Partial Compare Data Set

Order_ Delivery_
Obs Customer_ID Employee_ID Street_ID Date Date Order_ID

1 63 121039 9260125492 11JAN2003 11JAN2003 1230058123

2 41 120174 1600101527 28JAN2003 28JAN2003 1230147441
3 183 120134 1600100760 27FEB2003 27FEB2003 1230315085
4 56 121059 9260111871 15MAR2003 15MAR2003 1230404278
5 183 120149 1600100760 22MAR2003 22MAR2003 1230440481
6 183 120134 1600100760 25MAR2003 25MAR2003 1230455630
7 20 121066 9260118934 01APR2003 01APR2003 1230498538
8 79 121045 9260101874 18APR2003 18APR2003 1230591684

Median_
Total_Retail_ CostPrice_ Retail_
Obs Product_ID Quantity Price Per_Unit Discount Month Price

1 220101300017 1 $16.50 $7.45 . 1 $258.20

2 240600100010 2 $32.00 $6.50 . 1 $258.20
3 240200200039 3 $63.60 $8.80 . 2 $123.30
4 220200300002 2 $75.00 $17.05 . 3 $27.65
5 230100600005 1 $129.80 $63.20 . 3 $27.65
6 240200100233 2 $91.80 $22.45 . 3 $27.65
7 230100300006 1 $68.50 $34.35 . 4 $49.10
8 240200100076 4 $1,796.00 $246.55 . 4 $49.10

Level 2

2. Using a One-Dimensional Array as a Lookup Table

The data set orion.shoe_stats contains statistics for the shoe product lines.
Listing of orion.shoe_stats
orion.shoe_stats Data Set

Obs Stat Product21 Product22 Product23 Product24

1 Frequency 66.000 277.000 . 18.000

2 Mfg_Suggested_Retail_Price_Mean 70.788 174.292 . 173.056
3 Mfg_Suggested_Retail_Price_Min 17.000 13.000 . 5.000
4 Mfg_Suggested_Retail_Price_Max 130.000 385.000 . 398.000
5 Mfg_Suggested_Retail_Price_Median 68.000 164.000 . 190.500
6 Mfg_Suggested_Retail_Price_StdDev 21.731 71.703 . 141.389

a. Use arrays to create a data set named trans that has 24 observations.
5-20 Chapter 5 Using DATA Step Arrays

b. Print the trans data set.

PROC PRINT Output
The TRANS data set

Product_
Obs Stat Line Value

1 Frequency 21 66.000
2 Frequency 22 277.000
3 Frequency 23 .
4 Frequency 24 18.000
5 Mfg_Suggested_Retail_Price_Mean 21 70.788
6 Mfg_Suggested_Retail_Price_Mean 22 174.292
7 Mfg_Suggested_Retail_Price_Mean 23 .
8 Mfg_Suggested_Retail_Price_Mean 24 173.056
9 Mfg_Suggested_Retail_Price_Min 21 17.000
10 Mfg_Suggested_Retail_Price_Min 22 13.000
11 Mfg_Suggested_Retail_Price_Min 23 .
12 Mfg_Suggested_Retail_Price_Min 24 5.000
13 Mfg_Suggested_Retail_Price_Max 21 130.000
14 Mfg_Suggested_Retail_Price_Max 22 385.000
15 Mfg_Suggested_Retail_Price_Max 23 .
16 Mfg_Suggested_Retail_Price_Max 24 398.000
17 Mfg_Suggested_Retail_Price_Median 21 68.000
18 Mfg_Suggested_Retail_Price_Median 22 164.000
19 Mfg_Suggested_Retail_Price_Median 23 .
20 Mfg_Suggested_Retail_Price_Median 24 190.500
21 Mfg_Suggested_Retail_Price_StdDev 21 21.731
22 Mfg_Suggested_Retail_Price_StdDev 22 71.703
23 Mfg_Suggested_Retail_Price_StdDev 23 .
24 Mfg_Suggested_Retail_Price_StdDev 24 141.389

Level 3

3. Using a One-Dimensional Array

a. Use the program p305e03 to create a temporary data set named order_fact for the year 2007 and
customer IDs 89 and 2550, sorted by Order_Type.
The PROC SQL step creates a report showing the number of observations for each value of the
Order_Type variable for the Customer_ID values 89 and 2550.

Partial Listing of order_fact

order_fact

Order_ Order_ Delivery_

Obs Customer_ID Type Date Date Quantity

1 89 1 03JAN2007 04JAN2007 6
2 89 1 01OCT2007 01OCT2007 1
3 89 1 01OCT2007 01OCT2007 1
4 89 1 15DEC2007 15DEC2007 4
5 89 2 17JUN2007 21JUN2007 2
6 2550 3 04MAY2007 09MAY2007 3
7 2550 3 04MAY2007 09MAY2007 1
 5.1 Using One-Dimensional Arrays 5-21

b. Create the data set named all that has one observation for each Order_Type where there are a
varying number of observations for each Order_Type in the original data set order_fact. Use the
maximum number of observations for each order type as the array dimension to create three
arrays that create variables to hold the order dates, the delivery dates, and the quantity.
c. Print the first three observations of all.
PROC PRINT Output
The Resulting Data Set

Ordered_ Ordered_ Ordered_ Ordered_ Delivery_ Delivery_

Obs Date1 Date2 Date3 Date4 Date1 Date2

1 03JAN2007 01OCT2007 01OCT2007 15DEC2007 04JAN2007 01OCT2007

2 17JUN2007 . . . 21JUN2007 .
3 04MAY2007 04MAY2007 . . 09MAY2007 09MAY2007

Delivery_ Delivery_
Obs Date3 Date4 Quantity1 Quantity2 Quantity3 Quantity4

1 01OCT2007 15DEC2007 6 1 1 4
2 . . 2 . . .
3 . . 3 1 . .

Order_ Delivery_ Order_

Obs n Customer_ID Date Date Type Quantity

1 4 89 15DEC2007 15DEC2007 1 4
2 1 89 17JUN2007 21JUN2007 2 2
3 2 2550 04MAY2007 09MAY2007 3 1
5-22 Chapter 5 Using DATA Step Arrays

5.2 Using Multidimensional Arrays

Objectives
„ Define a multidimensional array.
„ Explain the differences between a one-dimensional
array and a multidimensional array.
„ Use a multidimensional array as a lookup table.

41

Business Scenario
The SAS data set orion.profit has information about
every company for the years 2003 through 2007,
separated by month.

Partial Listing of orion.profit(where=(Sales ne .))

Company YYMM Sales Cost Salaries Profit
Logistics 03M01 $457,809 $210,914 $127,525 $119,370
Logistics 03M02 $325,138 $149,718 $127,525 $47,895
Logistics 03M03 $276,805 $127,827 $134,198 $14,780
Logistics 03M04 $558,806 $264,868 $134,198 $159,741
Logistics 03M05 $641,954 $303,324 $134,198 $204,432
Logistics 03M06 $827,976 $389,207 $134,198 $304,571
Logistics 03M07 $819,373 $389,020 $138,047 $292,306
Logistics 03M08 $794,750 $373,204 $140,206 $281,340

42
 5.2 Using Multidimensional Arrays 5-23

5.05 Quiz
What is the type of the variable YYMM in the data set
orion.profit?

44

Business Scenario
This table contains the budgeted amounts for each of
those months and years. Each row represents a month,
and each column represents a year.
Yr2003 Yr2004 Yr2005 Yr2006 Yr2007
$1,590,000 $1,880,000 $2,300,000 $1,960,000 $1,970,000
$1,290,000 $1,550,000 $1,830,000 $1,480,000 $1,640,000
$1,160,000 $1,380,000 $1,640,000 $1,410,000 $1,440,000
$1,710,000 $2,100,000 $2,420,000 $2,130,000 $2,270,000
$1,990,000 $2,350,000 $2,840,000 $2,480,000 $2,670,000
$2,560,000 $3,020,000 $3,580,000 $3,070,000 $3,410,000
$2,590,000 $2,890,000 $3,550,000 $3,010,000 $3,490,000
$2,550,000 $2,840,000 $3,580,000 $3,030,000 $3,500,000
$1,070,000 $1,180,000 $1,550,000 $1,260,000 $1,520,000
$1,160,000 $1,270,000 $1,600,000 $1,360,000 $1,700,000
$1,260,000 $1,470,000 $1,780,000 $1,540,000 $1,950,000
$2,870,000 $3,120,000 $3,760,000 $3,210,000 $4,370,000

46 continued...

 The budget values in the table are not stored in a SAS data set.
5-24 Chapter 5 Using DATA Step Arrays

Business Scenario
You need to combine the budget amounts in the table
with the actual amount in the SAS data set to create the
following report:
Listing of budget_amt
Actual vs Budgeted Amounts (Two Observations)

Company YYMM Sales Cost Salaries Profit BudgetAmt

Logistics 03M01 $457,809 $210,914 $127,525 $119,370 $1,590,000

Logistics 03M02 $325,138 $149,718 $127,525 $47,895 $1,290,000

47

5.06 Quiz
What do the data set orion.profit and the lookup table
have in common?
Partial Listing of orion.profit (where=(Sales ne .))
Company YYMM Sales Cost Salaries Profit
Logistics 03M01 $457,809 $210,914 $127,525 $119,370
Logistics 03M02 $325,138 $149,718 $127,525 $47,895
Logistics 03M03 $276,805 $127,827 $134,198 $14,780
Logistics 03M04 $558,806 $264,868 $134,198 $159,741

Yr2003 Yr2004 Yr2005 Yr2006 Yr2007

$1,590,000 $1,880,000 $2,300,000 $1,960,000 $1,970,000
$1,290,000 $1,550,000 $1,830,000 $1,480,000 $1,640,000

49
 5.2 Using Multidimensional Arrays 5-25

Overview of Two-Dimensional Arrays

To combine the table of budgeted values with the data
set containing the profit, use a two-dimensional array.
A two-dimensional array is similar to a row of buckets.

1,1 1,2 2,1 2,2

„ SAS puts a value in a bucket based on multiple

numbers.
„ Values are retrieved from a bucket based on multiple
numbers.
51

Using Multidimensional Arrays

General form for the multidimensional ARRAY statement:
ARRAY array-name {…,rows, cols} <$> <length>
<elements> <(initial values)>;

rows specifies the number of array elements in a

row arrangement.
cols specifies the number of array elements in a
column arrangement.

Example: array B{2,5} B1-B10;

52

The keyword _TEMPORARY_ can be used instead of elements to avoid creating new variables in the
program data vector.
5-26 Chapter 5 Using DATA Step Arrays

Using Multidimensional Arrays

array B{2,5} B1-B10 (1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);

Yr2003 Yr2004 Yr2005 Yr2006 Yr2007

$1,590,000 $1,880,000 $2,300,000 $1,960,000 $1,970,000
$1,290,000 $1,550,000 $1,830,000 $1,480,000 $1,640,000
$1,160,000 $1,380,000 $1,640,000 $1,410,000 $1,440,000
$1,710,000 $2,100,000 $2,420,000 $2,130,000 $2,270,000
$1,990,000 $2,350,000 $2,840,000 $2,480,000 $2,670,000
$2,560,000 $3,020,000 $3,580,000 $3,070,000 $3,410,000
$2,590,000 $2,890,000 $3,550,000 $3,010,000 $3,490,000
$2,550,000 $2,840,000 $3,580,000 $3,030,000 $3,500,000
$1,070,000 $1,180,000 $1,550,000 $1,260,000 $1,520,000
$1,160,000 $1,270,000 $1,600,000 $1,360,000 $1,700,000
$1,260,000 $1,470,000 $1,780,000 $1,540,000 $1,950,000
$2,870,000 $3,120,000 $3,760,000 $3,210,000 $4,370,000

53

For this example, only the first two rows are included in the array.
The initial values fill all the columns in a row before moving on to the next row.

Using Multidimensional Arrays

array B{2,5} B1-B10 (1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);

PDV

B1 B2 B3 B4 B5 B6 B7 B8 B9 B10
1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

54

When you use a multidimensional array, the following statements are true:
• You must supply a subscript value for each dimension to process a specific array element.
• You can use a DO loop to process elements in a given dimension.
• You can use nested DO loops to process elements in more than one dimension.
 5.2 Using Multidimensional Arrays 5-27

5.07 Multiple Answer Poll

Which of the following would be equivalent to the following
ARRAY statement ?
array B{2,5} B1-B10 (1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);

a. array B{*} B1-B10

(1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);
b. array B{2,2003:2007} B1-B10
(1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);
c. array B{2,5} (1590000, 1880000, 2300000, 1960000,
1970000, 1290000, 1550000, 1830000,
1480000, 1640000);
d. array B{2,5} _temporary_ (1590000, 1880000, 2300000,
1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
56 1640000);

Business Scenario
Find the budgeted amounts for each company, year,
and month.

58
5-28 Chapter 5 Using DATA Step Arrays

Using Multidimensional Arrays

Find the budgeted amounts for each company, year,
and month.

Yr2003 Yr2004 Yr2005 Yr2006 Yr2007 Company YYMM

$1,590,000 $1,880,000 $2,300,000 $1,960,000 $1,970,000 Logistics 03M01

Logistics 03M02
$1,290,000 $1,550,000 $1,830,000 $1,480,000 $1,640,000

59

Using Multidimensional Arrays

data budget_amt;
drop Y M;
c array B{2,2003:2007} _temporary_
(1590000, 1880000, 2300000,
1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
1640000);
set orion.profit(where=(Sales ne .)
obs=2);
d Y=year(YYMM);
e M=month(YYMM);
f BudgetAmt=B{M,Y};
run;

p305d02
60

c Ten hardcoded values initialize the array. The _TEMPORARY_ keyword creates an array that is not
associated with variables in the program data vector.

d The variable Y (the column number) is calculated using the YEAR function on the date variable,
YYMM.

e The variable M (the row number) is created using the MONTH function on the date variable,
YYMM.

f The row and column numbers are used to look up the values of Budget in the array B.
 5.2 Using Multidimensional Arrays 5-29

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
. . . . . . . . 1

61 ...

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M01 457809 210914 127525 119370 . . . 1

62 ...
5-30 Chapter 5 Using DATA Step Arrays

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M01 457809 210914 127525 119370 2003 1 . 1

63 ...

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M01 457809 210914 127525 119370 2003 1 . 1

64 ...
 5.2 Using Multidimensional Arrays 5-31

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{1,2003};
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M01 457809 210914 127525 119370 2003 1 1590000 1

65 ...

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit Implicit
(1590000, OUTPUT;
1880000, 2300000,
Company YYMM Sales Cost . . . Implicit
1960000, RETURN;
1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M01 457809 210914 127525 119370 2003 1 1590000 1

66 ...
5-32 Chapter 5 Using DATA Step Arrays

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
Reinitialize PDV.
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M01 457809 210914 127525 119370 . . . 2

67 ...

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M02 325138 149718 127525 47895 2003 2 . 2

68 ...
 5.2 Using Multidimensional Arrays 5-33

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M02 325138 149718 127525 47895 2003 2 . 2

69 ...

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{2,2003};
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M02 325138 149718 127525 47895 2003 2 1290000 2

70 ...
5-34 Chapter 5 Using DATA Step Arrays

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit Implicit
(1590000, OUTPUT;
1880000, 2300000,
Company YYMM Sales Cost . . . Implicit
1960000, RETURN;
1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M02 325138 149718 127525 47895 2003 2 1290000 2

71 ...

Execution
data budget_amt;
drop Y M;
array B{2,2003:2007} _temporary_
Partial Listing of orion.profit (1590000, 1880000, 2300000,
Company YYMM Sales Cost . . . 1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
Logistics 03M01 457809 210914 . . . 1640000);
set orion.profit(where=(Sales ne .)
Logistics 03M02 325138 149718 . . . obs=2);
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
Execution stops. run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 1830000 1480000 1640000

PDV
Company YYMM Sales Cost Salaries Profit D Y DM BudgetAmt D_N_
Logistics 03M02 325138 149718 127525 47895 2003 2 1290000 2

72 ...
 5.2 Using Multidimensional Arrays 5-35

Exercises

Level 1

4. Using a Two-Dimensional Array

Orion Star wants to send discount coupons to their customers. The amounts of the discounts are given
in the following table:
Previous Quantity Ordered
OrderType 1 2 3 4 5 6
1 10 10 15 20 20 25
2 10 15 20 25 25 30
3 10 15 15 20 25 25

The data set orion.order_fact contains the variables Customer_ID, Quantity, and Order_Type.
Partial Listing of orion.order_fact
Order_
Obs Customer_ID Type Quantity

1 63 1 1
2 5 2 1
3 45 2 1
4 41 1 2
5 183 1 3
6 79 2 1
7 23 2 1
8 23 2 2
9 45 2 2
10 45 2 1

a. Use a two-dimensional array to combine the data set with the table of values to create a data set
named customer_coupons with a variable named Coupon_Value.
5-36 Chapter 5 Using DATA Step Arrays

b. Print the first five observations of the customer_coupons data set.

PROC PRINT Output
The Coupon Value

Order_ Coupon_
Obs Customer_ID Type Quantity Value

1 63 1 1 10
2 5 2 1 10
3 45 2 1 10
4 41 1 2 10
5 183 1 3 15

Level 2

5. Using a Two-Dimensional Array

The following table shows the average manufacturer’s suggested retail price for shoes, based on the
product line and product category:
Product Category

Product
Line 1 2

21 . 70.79

22 173.79 174.40

23 . .

24 29.65 287.8

The data set orion.shoe_sales contains the Product_ID, the Product_Name, and the
Total_Retail_Price for all of the shoes sold by Orion Star.
Partial Listing of orion.shoe_sales
Total_Retail_
Product_ID Product_Name Price

220200200024 Pro Fit Gel Gt 2030 Women's Running Shoes $178.50

220200100092 Big Guy Men's Air Terra Sebec Shoes $83.00
240200100043 Bretagne Performance Tg Men's Golf Shoes L. $282.40
220100700024 Armadillo Road Dmx Women's Running Shoes $99.70
220200300157 Hardcore Men's Street Shoes Large $220.20
240200100051 Bretagne Stabilites 2000 Goretex Shoes $420.90
220200100035 Big Guy Men's Air Deschutz Viii Shoes $125.20
220200100090 Big Guy Men's Air Terra Reach Shoes $177.20
220200200018 Lulu Men's Street Shoes $132.80
240200100052 Bretagne Stabilities Tg Men's Golf Shoes $99.70
 5.2 Using Multidimensional Arrays 5-37

a. Create a data set named combine using a two-dimensional array to combine the table of values
with the product line and the product category ID. The product line is the first two digits of the
Product_ID variable. The product category ID is the third and fourth digits of the Product_ID
variable.
b. Print the first five observations of the combine data set.
PROC PRINT Output
Total_Retail_
Obs Product_ID Product_Name Price

1 220200200024 Pro Fit Gel Gt 2030 Women's Running Shoes $178.50

2 220200100092 Big Guy Men's Air Terra Sebec Shoes $83.00
3 240200100043 Bretagne Performance Tg Men's Golf Shoes L. $282.40
4 220100700024 Armadillo Road Dmx Women's Running Shoes $99.70
5 220200300157 Hardcore Men's Street Shoes Large $220.20

Manufacturer_
Product_ Product_ Suggested_
Obs Prod_ID Line Cat_ID Price

1 220200200024 22 2 174.40
2 220200100092 22 2 174.40
3 240200100043 24 2 287.80
4 220100700024 22 1 173.79
5 220200300157 22 2 174.40

 The variables Product_Line and Product_Cat_ID must be numeric.

5-38 Chapter 5 Using DATA Step Arrays

Level 3
6. Using a Three-Dimensional Array
The warehouse location for the products in the orion.product_list data set is given in the following
table:
Warehouse Locations

Product_Line Product_Cat_ID Product_Loc_ID Warehouse

21 0 0 A2100

21 0 1 A2101

21 1 0 A2110

21 1 1 A2111

21 2 0 A2120

21 2 1 A2121

22 0 0 B2200

22 0 1 B2201

22 1 0 B2210

22 1 1 B2211

22 2 0 B2220

22 2 1 B2221

Open the program p305e06 that retrieves the Level 1products from the orion.product_list data set.
p305e06
data warehouses;
set orion.product_list(keep=Product_ID Product_Name
Product_Level
where=(Product_Level=1));
Prod_ID=put(Product_ID,12.);
Product_Line=input(substr(Prod_ID,1,2),2.);
Product_Cat_ID=input(substr(Prod_ID,3,2),2.);
Product_Loc_ID=input(substr(Prod_ID,12,1),1.);
/* subset the data for this exercise */
if Product_Line in (21,22) and Product_Cat_ID<=2
and Product_Loc_ID<2;
run;

Modify p305e06 to obtain the desired results.

 5.2 Using Multidimensional Arrays 5-39

a. Type the values of the Warehouse column into a three-dimensional array using the values of
Product_Line, Product_Cat_ID, and Product_Loc_ID as the dimensions.
b. Create a data set named warehouses. Use the Product_ID variable to determine the values of
Product_Line, Product_Cat_ID, and Product_Loc_ID.
• The product line is the first two digits of the Product_ID variable.
• The product category ID is the third and fourth digits of the Product_ID variable.
• The product location ID identifies the location within a warehouse of the product and is the last
digit of the Product_ID variable.
c. Print the first five observations of the warehouses data set.
PROC PRINT Output
Warehouses Data

Product_
Obs Product_ID Product_Name Level

1 210200400020 Kids Baby Edge Max Shoes 1

2 210200400070 Tony's Children's Deschutz (Bg) Shoes 1
3 210201000050 Kid Children's T-Shirt 1
4 220100100101 Big Guy Men's Chaser Poplin Pants 1
5 220100100241 Big Guy Men's Santos Shorts Dri Fit 1

Product_ Product_ Product_

Obs Prod_ID Line Cat_ID Loc_ID Warehouse

1 210200400020 21 2 0 A2120
2 210200400070 21 2 0 A2120
3 210201000050 21 2 0 A2120
4 220100100101 22 1 1 B2211
5 220100100241 22 1 1 B2211
5-40 Chapter 5 Using DATA Step Arrays

5.3 Loading a Multidimensional Array from a SAS Data Set

Objectives
„ Load a multidimensional array from a SAS data set.
„ Identify the advantages of an array as a lookup table.
„ Identify the disadvantages of an array as a lookup
table.

76

Business Scenario
Budget values are stored in a SAS data set named
orion.budget where the rows represent months
and the columns represent years.
Load the array from the values in the SAS data set.
Listing of orion.budget
Month Yr2003 Yr2004 Yr2005 Yr2006 Yr2007
1 $1,590,000 $1,880,000 $2,300,000 $1,960,000 $1,970,000
2 $1,290,000 $1,550,000 $1,830,000 $1,480,000 $1,640,000
3 $1,160,000 $1,380,000 $1,640,000 $1,410,000 $1,440,000
4 $1,710,000 $2,100,000 $2,420,000 $2,130,000 $2,270,000
5 $1,990,000 $2,350,000 $2,840,000 $2,480,000 $2,670,000
6 $2,560,000 $3,020,000 $3,580,000 $3,070,000 $3,410,000
7 $2,590,000 $2,890,000 $3,550,000 $3,010,000 $3,490,000
8 $2,550,000 $2,840,000 $3,580,000 $3,030,000 $3,500,000
9 $1,070,000 $1,180,000 $1,550,000 $1,260,000 $1,520,000
10 $1,160,000 $1,270,000 $1,600,000 $1,360,000 $1,700,000
11 $1,260,000 $1,470,000 $1,780,000 $1,540,000 $1,950,000
12 $2,870,000 $3,120,000 $3,760,000 $3,210,000 $4,370,000

77
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-41

Stored Array Values

Array values should be read from a SAS data set
when the following conditions exist:
„ There are too many values to initialize easily
in the array.
„ The values change frequently.

„ The same values are used in many programs.

78

Using Multidimensional Arrays

data budget_amt;
drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
if _N_=1 then do I=1 to 12; c
set orion.budget;
array tmp{2003:2007} Yr2003-Yr2007;
d do J=2003 to 2007;
e B{I,J}=tmp{J};
end;
end;
set orion.profit(where=(Sales ne .));
Y=year(YYMM);
M=month(YYMM);
BudgetAmt=B{M,Y};
run;
79 p305d03

The subscript variables I and J are used to process all the budget values in orion.budget.

c For each value of I, the SET statement reads an observation from the data set orion.budget
and fills a row in the tmp array.

d For each value of J, the yearly budget value, referenced through tmp{J}, is assigned to the
corresponding position J in the current row of the B array. The current row of the B array is
referenced by the value of I.

e The two-dimensional array B is loaded with the values of the tmp array.
5-42 Chapter 5 Using DATA Step Arrays

5.08 Multiple Choice Poll

How many elements are in the array defined
by the following ARRAY statement?

array B{12,2003:2007} _temporary_;

a. 0
b. 24
c. 48
d. 60

81

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

. . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. . . . . . .

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
83 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-43

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

. . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 . . . . . .

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
84 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

. . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 1 1590000 1880000 2300000 1960000 1970000 .

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
85 ...
5-44 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

. . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 1 1590000 1880000 2300000 1960000 1970000 2003

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
86 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

. . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 1 1590000 1880000 2300000 1960000 1970000 2003

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
87 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-45

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{1,2003}=tmp{2003};
B{I,J}= tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 1 1590000 1880000 2300000 1960000 1970000 2003

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
88 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 1 1590000 1880000 2300000 1960000 1970000 2004

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
89 ...
5-46 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
1 1 1590000 1880000 2300000 1960000 1970000 2004

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
90 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{1,2004}=tmp{2004};
B{I,J}= tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 1 1590000 1880000 2300000 1960000 1970000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
91 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-47

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
BudgetAmt=B{M,Y};
.
.
.
.
.
.
Continue
. . .
until J=2008.run;

1590000 1880000 2300000 1960000 1970000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 1 1590000 1880000 2300000 1960000 1970000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
92 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 1 1590000 1880000 2300000 1960000 1970000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
93 ...
5-48 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
94 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2003

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
95 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-49

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2003

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
96 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{2,2003}=tmp{2003};
B{I,J}= tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2003

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
97 ...
5-50 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2004

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
98 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{2,2004}=tmp{2004};
B{I,J}= tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2004

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
99 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-51

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
2 2 1290000 1550000 1830000 1480000 1640000 2005

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
100 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 Eventually,
1880000 . . . I=12 array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 and.J=2006
1550000 . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2006

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
101 ...
5-52 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . .

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2006

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
102 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{12,2006}=tmp{2006};
B{I,J}= tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2006

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
103 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-53

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2007

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
104 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2007

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
105 ...
5-54 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{12,2007}=tmp{2007};
B{I,J}= tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2007

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
106 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
12 12 2870000 3120000 3760000 3210000 4370000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
107 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-55

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
108 ...

Execution data budget_amt;

drop Yr2003-Yr2007 Month I J Y M;
Partial Listing of orion.budget array B{12,2003:2007} _temporary_;
Month Yr2003 Yr2004 . . . if _N_=1 then do I=1 to 12;
set orion.budget;
1 1590000 1880000 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
2 1290000 1550000 . . . B{I,J}=tmp{J};
end;
3 1160000 1380000 . . . end;
set orion.profit(where=(Sales ne .));
4 1710000 2100000 . . . Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
. . . . . . . . 1
109 ...
5-56 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008 Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 . . . 1
110 ...

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008 Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 2003 1 . 1
111 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-57

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008 Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 2003 1 . 1
112 ...

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
.
.
.
.
.
.
. . . BudgetAmt=B{1,2003};
BudgetAmt=B{M,Y};
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008 Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 2003 1 1590000 1
113 ...
5-58 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
end;
Logistics 03M03 276805 . . .
end; Implicit OUTPUT;
set orion.profit(where=(Sales ne .));
Logistics 03M04 558806 . . .
Y=year(YYMM); Implicit RETURN;
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
13 12 2870000 3120000 3760000 3210000 4370000 2008 Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 2003 1 1590000 1
114 ...

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . Reinitialize PDV.
run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. 12 2870000 3120000 3760000 3210000 4370000 . Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 . . . 2
115 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-59

Execution data budget_amt;

Partial ListingFalse
of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. 12 2870000 3120000 3760000 3210000 4370000 . Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M01 457809 210914 127525 119370 . . . 2
116 ...

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. 12 2870000 3120000 3760000 3210000 4370000 . Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M02 325138 149718 127525 47895 . . . 2
117 ...
5-60 Chapter 5 Using DATA Step Arrays

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. 12 2870000 3120000 3760000 3210000 4370000 . Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M02 325138 149718 127525 47895 2003 2 1290000 2
118 ...

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . .
Implicit
if _N_=1 then do I=1 to OUTPUT;
set orion.budget;
12;
Logistics 03M01 457809 . . .
Implicit RETURN;
array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 276805 . . .
end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. 12 2870000 3120000 3760000 3210000 4370000 . Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M02 325138 149718 127525 47895 2003 2 1290000 2
119 ...
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-61

Execution data budget_amt;

Partial Listing of orion.budget drop Yr2003-Yr2007 Month I J Y M;
array B{12,2003:2007} _temporary_;
Company YYMM Sales . . . if _N_=1 then do I=1 to 12;
set orion.budget;
Logistics 03M01 457809 . . . array tmp{2003:2007} Yr2003-Yr2007;
do J=2003 to 2007;
Logistics 03M02 325138 . . .
B{I,J}=tmp{J};
Logistics 03M03 Continue
276805. . . until EOF. end;
end;
Logistics 03M04 558806 . . . set orion.profit(where=(Sales ne .));
Y=year(YYMM);
. . . M=month(YYMM);
. . . . . . BudgetAmt=B{M,Y};
. . . run;

1590000 1880000 2300000 1960000 1970000 1290000 1550000 . . . 3210000 4370000

Partial PDV tmp{2003} tmp{2004} tmp{2005} tmp{2006} tmp{2007}

DI DMonth D Yr2003 DYr2004 D Yr2005 DYr2006 DYr2007 D J Company
. 12 2870000 3120000 3760000 3210000 4370000 . Logistics

Budget
YYMM Sales Cost Salaries Profit D Y DM D_N_
Amt
03M02 325138 149718 127525 47895 2003 2 1290000 2
120 ...

Using an Array
Advantages Disadvantages
of Using an Array of Using an Array
faster than a hash object or a contiguous chunk of memory
format if you can use it requested at compile time
use of positional order memory requirements to load
the entire array
use of multiple values to requirement that you must have
determine the array element to a numeric value as a pointer to
be returned the array elements
ability to use a non-sorted and the return of only a single value
non-indexed base data set from the lookup operation
use of numeric expressions to dimensions supplied at compile
determine which element of the time by either hardcoding or
array is to be looked up; exact macro variables
121
match not required
5-62 Chapter 5 Using DATA Step Arrays

Review of Arrays
Array

The subscript value(s) must be numeric.

One data value can be associated with the subscript value(s).

An array uses less memory than other in-memory lookup

techniques.
The size of the array is determined at compilation time.

Subscript values must be consecutive integers.

An array selects values by direct access based on the subscript

value.
Arrays can only be used in the DATA step.

122
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-63

Exercises

Level 1

7. Using a Two-Dimensional Array

Orion Star wants to send discount coupons to the customers. The amounts of the discounts are given
in the data set orion.coupons.
Listing of orion.coupons
orion.coupons Data Set

Obs OT Quantity1 Quantity2 Quantity3 Quantity4 Quantity5 Quantity6

1 1 10 10 15 20 20 25
2 2 10 15 20 25 25 30
3 3 10 15 15 20 25 25

The data set orion.order_fact contains variables Customer_ID, Order_Type, and Quantity.
Partial Listing of orion.order_fact
Order_
Obs Customer_ID Type Quantity

1 63 1 1
2 5 2 1
3 45 2 1
4 41 1 2
5 183 1 3
6 79 2 1
7 23 2 1
8 23 2 2
9 45 2 2
10 45 2 1

a. Create a two-dimensional array with the values from orion.coupons. Use values from
orion.order_fact and the array to create a new variable named Coupon_Value. Name the new
data set customer_coupons.
5-64 Chapter 5 Using DATA Step Arrays

b. Print the first 10 observations of the customer_coupons data set.

customer_coupons Data Set

Order_ Coupon_
Obs Customer_ID Type Quantity Value

1 63 1 1 10
2 5 2 1 10
3 45 2 1 10
4 41 1 2 10
5 183 1 3 15
6 79 2 1 10
7 23 2 1 10
8 23 2 2 15
9 45 2 2 15
10 45 2 1 10

8. Using a Two-Dimensional Array (Optional)

Orion Star wants to send discount coupons to the customers. The amounts of the discounts are given
in the data set orion.coupon_pct.
Listing of orion.coupon_pct
orion.coupon_pct

Obs OT Quant Value

1 1 1 10
2 1 2 10
3 1 3 15
4 1 4 20
5 1 5 20
6 1 6 25
7 2 1 10
8 2 2 15
9 2 3 20
10 2 4 25
11 2 5 25
12 2 6 30
13 3 1 10
14 3 2 15
15 3 3 15
16 3 4 20
17 3 5 25
18 3 6 25

The data set orion.order_fact contains variables Customer_ID, Order_Type, and Quantity.
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-65

Partial Listing of orion.order_fact

Order_
Obs Customer_ID Type Quantity

1 63 1 1
2 5 2 1
3 45 2 1
4 41 1 2
5 183 1 3
6 79 2 1
7 23 2 1
8 23 2 2
9 45 2 2
10 45 2 1

a. Create a two-dimensional array with the values from orion.coupon_pct. Use values from
orion.order_fact and the array to create a new variable named Coupon_Value. Name the new
data set customer_coupons.
b. Print the first 10 observations of the customer_coupons data set.
PROC PRINT Output
The Coupon Value

Order_ Coupon_
Obs Customer_ID Type Quantity Value

1 63 1 1 10
2 5 2 1 10
3 45 2 1 10
4 41 1 2 10
5 183 1 3 15
6 79 2 1 10
7 23 2 1 10
8 23 2 2 15
9 45 2 2 15
10 45 2 1 10
5-66 Chapter 5 Using DATA Step Arrays

Level 2

9. Using a Two-Dimensional Array

The data set orion.msp contains the average manufacturer’s suggested retail price for shoes, based on
the product line and the product category. The product group ID is the last two digits of
Prod_Cat_ID.
Listing of orion.msp
orion.msp

Prod_ Prod_ Avg_Suggested_

Obs Line Cat_ID Retail_Price

1 21 2101 .
2 21 2102 70.79
3 22 2201 173.79
4 22 2202 174.40
5 23 2301 .
6 23 2302 .
7 24 2401 29.63
8 24 2402 287.80

The data set orion.shoe_sales contains the Product_ID, Product_Name, and Total_Retail_Price
for all of the shoes sold by Orion Star.
Partial Listing of orion.shoe_sales
Total_Retail_
Product_ID Product_Name Price

220200200024 Pro Fit Gel Gt 2030 Women's Running Shoes $178.50

220200100092 Big Guy Men's Air Terra Sebec Shoes $83.00
240200100043 Bretagne Performance Tg Men's Golf Shoes L. $282.40
220100700024 Armadillo Road Dmx Women's Running Shoes $99.70
220200300157 Hardcore Men's Street Shoes Large $220.20
240200100051 Bretagne Stabilites 2000 Goretex Shoes $420.90
220200100035 Big Guy Men's Air Deschutz Viii Shoes $125.20
220200100090 Big Guy Men's Air Terra Reach Shoes $177.20
220200200018 Lulu Men's Street Shoes $132.80
240200100052 Bretagne Stabilities Tg Men's Golf Shoes $99.70

a. Create a data set named combine using a two-dimensional array to combine the table of values from
orion.msp with orion.shoe_sales. Create a new variable named Manufacturer_Suggested_Price
based on the values of product line and product category. The product line is the first two digits of
the Product_ID variable. The product category ID is the third and fourth digits of the Product_ID
variable. Keep only the Product_ID, Product_Name, Total_Retail_Price, and
Manufacturer_Suggested_Price variables.
 5.3 Loading a Multidimensional Array from a SAS Data Set 5-67

b. Print the first five observations of the combine data set.

PROC PRINT Output
Manufacturer_
Suggested_ Total_Retail_
Obs Price Product_ID Product_Name Price

1 $174.40 220200200024 Pro Fit Gel Gt 2030 Women's Running Shoes $178.50
2 $174.40 220200100092 Big Guy Men's Air Terra Sebec Shoes $83.00
3 $287.80 240200100043 Bretagne Performance Tg Men's Golf Shoes L. $282.40
4 $173.79 220100700024 Armadillo Road Dmx Women's Running Shoes $99.70
5 $174.40 220200300157 Hardcore Men's Street Shoes Large $220.20

Level 3

10. Using a Three-Dimensional Array

The data set orion.warehouses contains the warehouse location for all the products. The
Product_Line variable has values from 21 to 24 inclusive, the Product_Cat_ID variable has values
from 0 to 8 inclusive, and the variable Product_Loc_ID has values from 0 to 9 inclusive.

Partial Listing of orion.warehouses

orion.warehouses (Partial Data Set)

Product_ Product_ Product_

Obs Line Cat_ID Loc_ID Warehouse

1 21 0 0 A2100
2 21 0 1 A2101
3 21 1 0 A2110
4 21 1 1 A2111
5 21 2 0 A2120
6 21 2 2 A2122
7 21 2 3 A2123
8 21 2 4 A2124
9 21 2 5 A2125
10 21 2 6 A2126

a. Write a DATA step to create a data set named warehouses.

b. Load the values from orion.warehouses into a three-dimensional array.
c. Read only the variables Product_ID, Product_Name, and Product_Level and all of the
observations from orion.product_list where Product_Level = 1. Use the Product_ID variable
to determine the values of Product_Line, Product_Cat_ID, and Product_Loc_ID.
• The product line is the first two digits of the Product_ID variable.
• The product category ID is the third and fourth digits of the Product_ID variable.
• The product location ID identifies the location within a warehouse of the product and is the last
digit of the Product_ID variable.
d. Use Product_Line, Product_Cat_ID, and Product_Loc_ID to retrieve the value from the array
and create a variable named Warehouse.
e. Keep only the variables Product_ID, Product_Name, and Warehouse.
5-68 Chapter 5 Using DATA Step Arrays

f. Print the first five observations of the warehouses data set.

PROC PRINT Output
warehouses

Obs Warehouse Product_ID Product_Name

1 A2129 210200100009 Kids Sweat Round Neck,Large Logo

2 A2127 210200100017 Sweatshirt Children's O-Neck
3 A2122 210200200022 Sunfit Slow Swimming Trunks
4 A2123 210200200023 Sunfit Stockton Swimming Trunks Jr.
5 A2126 210200300006 Fleece Cuff Pant Kid'S
 5.4 Chapter Review 5-69

5.4 Chapter Review

Chapter Review
1. Define an array.

2. When is an array deleted from memory?

3. How can you visualize a two-dimensional array?

125

Chapter Review
4. How many elements are created in the following
ARRAY statement?
array myarray{5:9,7};

5. What are the names of the variables created

by the ARRAY statement in question 4?

127
5-70 Chapter 5 Using DATA Step Arrays

5.5 Solutions

Solutions to Exercises
1. Using a One-Dimensional Array to Combine Data
a. Combine the two data sets to create a data set named compare. The data set should contain the
variables from orion.retail and variables named Month and Median_Retail_Price, where
Month is the month of the date that the product was ordered.
b. Print the first eight observations of the resulting data set.
p305s01
data compare;
drop Month1-Month12 Statistic;
array mon{12} Month1-Month12;
if _N_=1 then
set orion.retail_information
(where=(Statistic='Median_Retail_Price'));
set orion.retail;
Month=month(Order_Date);
Median_Retail_Price=mon(Month);
run;

proc print data=compare(obs=8);

title 'Partial Compare Data Set';
run;
2. Using a One-Dimensional Array as a Lookup Table
a. Use arrays to create a data set named trans that has 24 observations.
b. Print the trans data set.
p305s02
data trans;
drop Product21-Product24;
array prod{21:24} Product21-Product24;
set orion.shoe_stats;
do Product_Line=21 to 24;
Value=prod{Product_Line};
output;
end;
run;

proc print data=trans;

title 'The TRANS data set';
run;
 5.5 Solutions 5-71

3. Using a One-Dimensional Array

a. Use the program p305e03 to create a temporary data set order_fact for the year 2007 and
customer IDs 89 and 2550, sorted by Order_Type.
b. Create the data set named all that has one observation for each Order_Type where there are a
varying number of observations for each Order_Type in the original data set order_fact. Use the
maximum number of observations for each order type as the array dimension to create three
arrays that create variables to hold the order dates, the delivery dates, and the quantity.
c. Print the first three observations of all.
p305s03
/************************************************************/
/* The SORT step limits the data to a few rows for practice */
/* and sorts the data to use with a BY statement in the */
/* DATA step. */
/************************************************************/

proc sort data=orion.order_fact out=order_fact(keep=Customer_ID

Order_Type Order_Date Delivery_Date Quantity);
where Customer_ID in (89, 2550) and year(Order_Date)=2007;
by Order_Type;
run;

/************************************************************/
/* The SQL step counts the number of Order Types so that */
/* you know the dimensions for the arrays that the program */
/* needs. */
/************************************************************/

proc sql;
select Order_Type, count(*)
from order_fact
group by Order_Type;
quit;
(Continued on the next page.)
5-72 Chapter 5 Using DATA Step Arrays

/************************************************************/
/* The DATA step creates 4 variables for the order dates, */
/* 4 for the delivery dates, and 4 for the quantities. */
/* N is a counter of observations for each Order_Type. */
/* N needs to initialized to 0 when the DATA step iterates. */
/* The DATA step will execute a new time when the DO UNTIL */
/* loop ends. This happens when the last observation */
/* for an Order_Type has been processed. */
/* The three assignment statements in the DO loop */
/* are creating the variables for each value of Order_Type. */
/************************************************************/

data all;
array ordt{*} Ordered_Date1-Ordered_Date4;
array deldt{*} Delivery_Date1-Delivery_Date4;
array q{*} Quantity1 - Quantity4;
format Ordered_Date1-Ordered_Date4
Delivery_Date1-Delivery_Date4
date9.;
N=0;
do until (last.Order_Type);
set order_fact;
by Order_Type;
N+1;
ordt{N}=Order_Date;
deldt{N}=Delivery_Date;
q{N}=Quantity;
end;
run;

proc print data=all;

run;
(Continued on the next page.)
 5.5 Solutions 5-73

/***********************************************************/
/* to get a macro variable for the number of observations */
/* in order_fact */
/* */
/* proc sql; */
/* create table temp as */
/* select count(*) as Num */
/* from order_fact */
/* group by Customer_ID; */
/* select max(num) into :NumObs */
/* from temp; */
/* */
/* Then substitute &NumObs into the program instead of */
/* the 4 */
/***********************************************************/

proc sort data=orion.order_fact out=order_fact(keep=Customer_ID

Order_Type Order_Date Delivery_Date Quantity);
where Customer_ID in (89, 2550) and year(Order_Date)=2007;
by Order_Type;
run;

proc sql;
create table temp as
select count(*) as Num
from order_fact
group by Order_Type
select max(num) into :NumObs
from temp;
%let NumObs=&NumObs;
quit;

data all;
array ordt{*} Ordered_Date1-Ordered_Date&NumObs;
array deldt{*} Delivery_Date1-Delivery_Date&NumObs;
array q{*} Quantity1 - Quantity&NumObs;
format Ordered_Date1-Ordered_Date&NumObs
Delivery_Date1-Delivery_Date&NumObs
date9.;
N=0;
do until (last.Order_Type);
set order_fact;
by Order_Type;
N+1;
ordt{N}=Order_Date;
deldt{N}=Delivery_Date;
q{N}=Quantity;
end;
run;
proc print data=all;
run;
5-74 Chapter 5 Using DATA Step Arrays

4. Using a Two-Dimensional Array

a. Use a two-dimensional array to combine the data set with the table of values to create a data set
named customer_coupons with a variable named Coupon_Value.
b. Print the first five observations of the customer_coupons data set.
p305s04
data customer_coupons;
array pct{3,6} _temporary_ (10, 10, 15, 20, 20, 25,
10, 15, 20, 25, 25, 30,
10, 15, 15, 20, 25, 25);
set orion.order_fact(keep=Customer_ID Order_Type Quantity);
Coupon_Value=pct(Order_Type,Quantity);
run;

proc print data=customer_coupons(obs=5);

title 'The Coupon Value';
run;
5. Using a Two-Dimensional Array
a. Create a data set named combine using a two-dimensional array to combine the table of values
with the product line and the product category ID.
b. Print the first five observations of the combine data set.
p305s05
data combine;
array msp{21:24,2} _temporary_
(., 70.79, 173.79, 174.40, ., .,29.65, 287.8);
set orion.shoe_sales;
Prod_ID=put(Product_ID,12.);
Product_Line=input(substr(Prod_ID,1,2),2.);
Product_Cat_ID=input(substr(Prod_ID,3,2),2.);
Manufacturer_Suggested_Price=msp{Product_Line,
Product_Cat_ID};
run;

proc print data=combine(obs=5);

run;
6. Using a Three-Dimensional Array
Open the program p305e06 that retrieves the Level 1products from the orion.product_list data set.
Modify p305e06 to obtain the desired results.
a. Type the values of the Warehouse column into a three-dimensional array using the values of
Product_Line, Product_Grp_ID, and Product_Cat_ID as the dimensions.
 5.5 Solutions 5-75

b. Create a data set named warehouses. Use the Product_ID variable to determine the values of
Product_Line, Product_Cat_ID, and Product_Loc_ID.
• The product line is the first two digits of the Product_ID variable.
• The product category ID is the third and fourth digits of the Product_ID variable.
• The product location ID identifies the location within a warehouse of the product and is the last
digit of the Product_ID variable.
c. Print the first five observations of the warehouses data set.
p305s06
data warehouses;
array W{21:22,0:2,0:1} $ 5 _temporary_ ('A2100',
'A2101',
'A2110',
'A2111',
'A2120',
'A2121',
'B2200',
'B2201',
'B2210',
'B2211',
'B2220',
'B2221');
set orion.product_list(keep=Product_ID Product_Name
Product_Level
where=(Product_Level=1));
Prod_ID=put(Product_ID,12.);
Product_Line=input(substr(Prod_ID,1,2),2.);
Product_Cat_ID=input(substr(Prod_ID,3,2),2.);
Product_Loc_ID=input(substr(Prod_ID,12,1),1.);
/* subset the data for this exercise */
if Product_Line in (21,22) and Product_Cat_ID<=2
and Product_Loc_ID<2;
Warehouse=W(Product_Line, Product_Cat_ID, Product_Loc_ID);
run;

proc print data=warehouses(obs=5);

title 'Warehouses Data';
run;
7. Using a Two-Dimensional Array
a. Create a two-dimensional array with the values from orion.coupons. Use values from
orion.order_fact and the array to create a new variable named Coupon_Value. Name the new
data set customer_coupons.
5-76 Chapter 5 Using DATA Step Arrays

b. Print the first 10 observations of the customer_coupons data set.

p305s07
data customer_coupons;
drop ot i j quantity1-quantity6;
array pct{3,6} _temporary_;
if _n_=1 then do i=1 to 3;
set orion.coupons;
array quan{6} Quantity1-Quantity6;
do j=1 to 6;
pct{i,j}=quan{j};
end;
end;
set orion.order_fact(keep=Customer_ID Order_Type Quantity);
Coupon_Value=pct{Order_Type,Quantity};
run;

proc print data=customer_coupons(obs=10);

title 'customer_coupons Data Set';
run;
8. Using a Two-Dimensional Array (Optional)
a. Create a two-dimensional array with the values from orion.coupon_pct. Use values from
orion.order_fact and the array to create a new variable named Coupon_Value. Name the new
data set customer_coupons.
b. Print the first 10 observations of the customer_coupons data set.
p305s08
data customer_coupons;
drop OT Quant Value i;
array pct{3,6} _temporary_ ;
if _N_=1 then do i=1 to All;
set orion.coupon_pct nobs=All;
pct{OT,Quant}=Value;
end;
set orion.order_fact(keep=Customer_ID Order_Type Quantity);
Coupon_Value=pct(Order_Type,Quantity);
run;
proc print data=customer_coupons(obs=10);
title 'The Coupon Value';
run;
9. Using a Two-Dimensional Array
a. Create a data set named combine using a two-dimensional array to combine the table of values from
orion.msp with orion.shoe_sales. Create a new variable named Manufacturer_Suggested_Price
based on the values of product line and product category. The product line is the first two digits of
the Product_ID variable. The product category ID is the third and fourth digits of the Product_ID
variable. Keep only the Product_ID, Product_Name, Total_Retail_Price, and
Manufacturer_Suggested_Price variables.
 5.5 Solutions 5-77

b. Print the first five observations of the combine data set.

p305s09
data combine;
array msp{21:24,2} _temporary_ ;
keep Product_ID Product_Name Total_Retail_Price
Manufacturer_Suggested_Price;
format Manufacturer_Suggested_Price dollar8.2;
if _N_= 1 then do i=1 to All;
set orion.msp nobs=All;
msp{Prod_Line,input(substr(put(Prod_Cat_ID,4.),3,2),2.)}
=Avg_Suggested_Retail_Price;
end;
set orion.shoe_sales;
Prod_ID=put(Product_ID,12.);
Product_Line=input(substr(Prod_ID,1,2),2.);
Product_Cat_ID=input(substr(Prod_ID,3,2),2.);
Manufacturer_Suggested_Price=
msp{Product_Line, Product_Cat_ID};
run;
proc print data=combine(obs=5);
run;
10. Using a Three-Dimensional Array
a. Write a DATA step to create a data set named warehouses.
b. Load the values from orion.warehouses into a three-dimensional array.
c. Read only the variables Product_ID, Product_Name, and Product_Level and all of the
observations from orion.product_list where Product_Level = 1. Use the Product_ID variable
to determine the values of Product_Line, Product_Cat_ID, and Product_Loc_ID.
• The product line is the first two digits of the Product_ID variable.
• The product category ID is the third and fourth digits of the Product_ID variable.
• The product location ID identifies the location within a warehouse of the product and is the last
digit of the Product_ID variable.
d. Use Product_Line, Product_Cat_ID, and Product_Loc_ID to retrieve the value from the array
and create a variable named Warehouse.
e. Keep only the variables Product_ID, Product_Name, and Warehouse.
5-78 Chapter 5 Using DATA Step Arrays

f. Print the first five observations of the warehouses data set.

p305s10
data warehouses;
keep Product_ID Product_Name Warehouse;
array w{21:24,0:8,0:9} $ 5 _temporary_ ;
if _n_=1 then do i=1 to all;
set orion.warehouses nobs=all;
W{Product_Line, Product_Cat_ID, Product_Loc_ID}=Warehouse;
end;
set orion.product_list(keep=Product_ID Product_Name
Product_Level
where=(Product_Level=1));
Prod_ID=put(Product_ID,12.);
Product_Line=input(substr(Prod_ID,1,2),2.);
Product_Cat_ID=input(substr(Prod_ID,3,2),2.);
Product_Loc_ID=input(substr(Prod_ID,12,1),1.);
Warehouse=w(Product_Line, Product_Cat_ID, Product_Loc_ID);
run;

proc print data=warehouses(obs=5);

title 'warehouses';
run;
 5.5 Solutions 5-79

Solutions to Student Activities (Polls/Quizzes)

5.01 Multiple Choice Poll – Correct Answer

How many elements are referenced by the following
ARRAY statement?

array numarray{*} Num1 – Num12;

a. 0
b. 1
c. 12
d. Unknown

13

5.02 Poll – Correct Answer

Can the two data sets be merged with the DATA step
MERGE statement or joined with the SQL procedure
without pre-processing the data?
€ Yes
€ No

21
5-80 Chapter 5 Using DATA Step Arrays

5.03 Poll – Correct Answer

What do the two data sets have in common?
€ They have the year in common.
€ They have nothing in common.
In the data set orion.salary_stats, the columns,
except for the first, represent the year values.
In the data set orion.employee_payroll,
the year values can be obtained from the
Employee_Hire_Date variable.

23

5.04 Multiple Answer Poll – Correct Answers

Which of the following ARRAY statements are similar
to the statement
array yr{1974:2007} Yr1974-Yr2007;

and will compile without errors?

a. array yr{34} Yr1974-Yr2007;
b. array yr{1974-2007} Yr1974-Yr2007;
c. array yr{74:07} Yr1974-Yr2007;
d. array yr{74-07} Yr1974-Yr2007;
e. array yr{*} Yr1974-Yr2007;

37
 5.5 Solutions 5-81

5.05 Quiz – Correct Answer

What is the type of the variable YYMM in the data set
orion.profit?
Use PROC CONTENTS.

proc contents data=orion.profit;

run;

YYMM is a numeric variable. It represents a SAS date.

45

5.06 Quiz – Correct Answer

What do the data set orion.profit and the lookup table
have in common?

They have the month and year in common. In the data

set orion.profit, the month and year can be extracted
from the variable YYMM. In the table, each row
represents a month and each column represents a
year.

50
5-82 Chapter 5 Using DATA Step Arrays

5.07 Multiple Answer Poll – Correct Answers

Which of the following would be equivalent to the following
ARRAY statement ?
array B{2,5} B1-B10 (1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);

a. array B{*} B1-B10

(1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);
b. array B{2,2003:2007} B1-B10
(1590000, 1880000, 2300000, 1960000, 1970000,
1290000, 1550000, 1830000, 1480000, 1640000);
c. array B{2,5} (1590000, 1880000, 2300000, 1960000,
1970000, 1290000, 1550000, 1830000,
1480000, 1640000);
d. array B{2,5} _temporary_ (1590000, 1880000, 2300000,
1960000, 1970000, 1290000,
1550000, 1830000, 1480000,
57 1640000);

5.08 Multiple Choice Poll – Correct Answer

How many elements are in the array defined
by the following ARRAY statement?

array B{12,2003:2007} _temporary_;

a. 0
b. 24
c. 48
d. 60

82
 5.5 Solutions 5-83

Solutions to Chapter Review

Chapter Review – Correct Answers

1. Define an array.
An array is a temporary grouping of SAS variables
that are arranged in a particular order and
identified by an array name.
2. When is an array deleted from memory?
An array is deleted when the DATA step completes
execution.
3. How can you visualize a two-dimensional array?
As a table having a certain number of rows
and a certain number of columns

126

Chapter Review – Correct Answers

4. How many elements are created in the following
ARRAY statement?
array myarray{5:9,7};

35
5. What are the names of the variables created
by the ARRAY statement in question 4?
myarray1 – myarray35

128
5-84 Chapter 5 Using DATA Step Arrays
Chapter 6 Using DATA Step Hash
and Hiter Objects

6.1 Introduction..................................................................................................................... 6-3

6.2 Using Hash Object Methods .......................................................................................... 6-7

Exercises .............................................................................................................................. 6-28

6.3 Loading a Hash Object with Data from a SAS Data Set ............................................ 6-31
Exercises .............................................................................................................................. 6-42

6.4 Using the DATA Step Hiter Object ............................................................................... 6-48

Exercises .............................................................................................................................. 6-65

6.5 Using a Hash Object for Chained Lookups (Self-Study) ........................................... 6-67
Demonstration: Creating a List of Values............................................................................. 6-82

Exercises .............................................................................................................................. 6-84

6.6 Chapter Review............................................................................................................. 6-87

6.7 Solutions ....................................................................................................................... 6-89

Solutions to Exercises .......................................................................................................... 6-89

Solutions to Student Activities (Polls/Quizzes) ................................................................... 6-100

Solutions to Chapter Review .............................................................................................. 6-105

6-2 Chapter 6 Using DATA Step Hash and Hiter Objects
 6.1 Introduction 6-3

6.1 Introduction

Objectives
„ Define the DATA step hash object.

6.01 Poll
Have you used hash objects in SAS or other computer
languages?
€ Yes
€ No

5
6-4 Chapter 6 Using DATA Step Hash and Hiter Objects

DATA Step Hash Objects

The DATA step hash object has the following attributes:
„ provides in-memory data storage and retrieval

„ has a data component and a key component

„ uses the key for quick data retrieval

„ can store multiple data items per key

„ does not require the data to be sorted

„ is sized dynamically

 The hash object is a good choice for lookups using

unordered data that can fit into memory.

 Additional information about the hash object is available at the DATA Step Community Web site:
support.sas.com/rnd/base/index-datastep.html

Overview of a Hash Object (Review)

A hash object is similar to rows of buckets that are
identified by the value of a key.
Key Data Data
„ SAS puts value(s)
in the data bucket(s)
based on the
value(s) in the key
bucket.
„ Value(s) are
retrieved from the
data bucket(s)
based on the
value(s) in the key
bucket.
7
 6.1 Introduction 6-5

DATA Step Hash Objects

The hash object resembles a table with rows and
columns. The columns have the following characteristics:
„ can be numeric or character

„ can be loaded from hardcoded values

„ can be loaded from a SAS data set

„ exist for the duration of the DATA step

„ can be output to a SAS data set

DATA Step Hash Objects

The key component has the following attributes:
„ can consist of numeric and character values

„ maps key values to data rows

„ must be unique before SAS 9.2

„ can be composite

The data component has the following attributes:

„ can contain multiple data values per key value

„ can consist of numeric and character values

Data components and key components are

DATA step variables.

9
6-6 Chapter 6 Using DATA Step Hash and Hiter Objects

Using Hash Objects

The DATA step hash object has these characteristics:
„ is created with a DECLARE statement

„ has attributes and methods

„ is manipulated with object dot syntax

An attribute is a property.
A method is a function.

10
 6.2 Using Hash Object Methods 6-7

6.2 Using Hash Object Methods

Objectives
„ Investigate hash object syntax.
„ Use hash object methods to load data into
a hash object.
„ Use a hash object method to match records.

12

Business Scenario
The SAS data set orion.europe_customers has
variables that contain the customer type for the last year
and for the current year.
Listing of orion.europe_customers
LastYr ThisYr
Customer_Name Customer_Address Country
Type Type
Cornelia Krahl Kallstadterstr. 9 DE 20 20
Elke Wallstab Carl-Zeiss-Str. 15 DE 10 20
Markus Sepke Iese 1 DE 20 10
Ulrich Heyde Oberstr. 61 DE 30 10
Oliver S. Füßling Hechtsheimerstr. 18 DE 20 30
Rolf Robak Münsterstraße 67 DE 10 30
Thomas Leitmann Carl Von Linde Str. 13 DE 10 20
Gert-Gunter Mendler Humboldtstr. 1 DE 20 30
Carsten Maestrini Münzstr. 28 DE 20 30
Ines Deisser Bahnweg 1 DE 10 20

13
6-8 Chapter 6 Using DATA Step Hash and Hiter Objects

Business Scenario
Code Member Type
Customer descriptions must be assigned Orion Club
based on customer code values for 10
members
member type. The values are shown Orion Club
20
Gold members
in the table on the right but are not stored Internet/
in a SAS data set. 30 Catalog
customers
Listing of orion.europe_customers
LastYr ThisYr
Customer_Name Customer_Address Country
Type Type
Cornelia Krahl Kallstadterstr. 9 DE 20 20
Elke Wallstab Carl-Zeiss-Str. 15 DE 10 20
Markus Sepke Iese 1 DE 20 10
Ulrich Heyde Oberstr. 61 DE 30 10
Oliver S. Füßling Hechtsheimerstr. 18 DE 20 30
Rolf Robak Münsterstraße 67 DE 10 30
Thomas Leitmann Carl Von Linde Str. 13 DE 10 20
Gert-Gunter Mendler Humboldtstr. 1 DE 20 30
Carsten Maestrini Münzstr. 28 DE 20 30
Ines Deisser Bahnweg 1 DE 10 20
14

A set of lookup values can be stored in a hash object. Whereas an array uses a series of consecutive
integers to address array elements, a hash object can use any combination of numeric and character values
as addresses.

6.02 Multiple Answer Poll

Which of the following could be used to assign
descriptions to each of the variables ThisYrType
and LastYrType?
a. Merging
b. Formats
c. IF-THEN/ELSE
d. Arrays

16
 6.2 Using Hash Object Methods 6-9

Using Hash Objects

Load the code and member type into a hash object
named T.

Code MemberType

HASH Object T
KEY DATA
10 Orion Club members
20 Orion Club Gold members
30 Internet/Catalog Customers

18

Using Hash Objects

data mem_type;
length Code $2 MemberType $40;
if _N_=1 then do;
declare hash T();
T.definekey('Code');
T.definedata('MemberType');
T.definedone();
T.add(key:'10', data:'Orion Club members');
T.add(key:'20', data:'Orion Club Gold members');
T.add(key:'30', data:'Internet/Catalog Customers');
end;
set orion.europe_customers;
rc1=T.find(key:ThisYrType);
if rc1=0 then ThisYrMember=MemberType;
rc2=T.find(key:LastYrType);
if rc2=0 then LastYrMember=MemberType;
run;

p306d01
19
6-10 Chapter 6 Using DATA Step Hash and Hiter Objects

Compilation
data mem_type;
length Code $2 MemberType $40;
if _N_=1 then do;
declare hash T();
T.definekey('Code');
T.definedata('MemberType');
T.definedone();
T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
T.add(key:'30',data:'Internet/Catalog Customers');
end;
set orion.europe_customers;
rc1=T.find(key:ThisYrType);
if rc1=0 then ThisYrMember=MemberType;
rc2=T.find(key:LastYrType);
if rc2=0 then LastYrMember=MemberType;
run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . .

20 ...

Execution
Partial orion.europe_customers
True
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
Markus
T.definedata('MemberType');
. . 20 10 T.definedone();
Sepke
. . . T.add(key:'10',data:'Orion Club members');
. . . . . T.add(key:'20',data:'Orion Club Gold members');
. . . T.add(key:'30',data:'Internet/Catalog Customers');
end;
set orion.europe_customers;
rc1=T.find(key:ThisYrType);
if rc1=0 then ThisYrMember=MemberType;
rc2=T.find(key:LastYrType);
if rc2=0 then LastYrMember=MemberType;
run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . 1

21 ...
 6.2 Using Hash Object Methods 6-11

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
Gold members rc2=T.find(key:LastYrType);
Internet/Catal if rc2=0 then LastYrMember=MemberType;
30
og Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . 1

22 ...

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . 1

25 ...
6-12 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . 1

29 ...

6.03 Multiple Answer Poll

When would you not use quotation marks around the
value for an argument to a method?
a. The value is numeric.
b. The value is character.
c. The value of a PDV variable is wanted.
d. The value is character or numeric.

31
 6.2 Using Hash Object Methods 6-13

Declaring a Hash Object

declare hash T();

General form for the DECLARE statement:

DECLARE object object-reference (<arg_tag-1: value-1

<,…arg_tag-n: value-n>>);

object specifies the component object.

object-reference specifies the object-reference name for the
component object.
arg_tag specifies the information that is used to
create an instance of the component object.
value specifies the value for an argument tag.
33

Valid values for arg_tag depend on the component object.

 When a DATA step hash object is created, it is said to be instantiated.

Declaring a Hash Object

Valid values for object are as follows:

hash indicates a hash object.

hiter indicates a hash iterator object.

The hiter object retrieves data from the hash object in

ascending or descending key order.

34
6-14 Chapter 6 Using DATA Step Hash and Hiter Objects

Hash Object Argument Tags

Argument_tag Value Description
dataset: 'data-set_name' name of a SAS data set to load
into the hash object
hashexp: n hash object's table size, where
the size of the hash table is 2n
(default n=8, max n=16)
ordered: 'NO' | sort order for the OUTPUT
'ascending' | method or the hash iterator
'descending' | object (default='NO')
'YES' | 'Y'

35
 6.2 Using Hash Object Methods 6-15

Reference Information

HASHEXP
In order to maximize the efficiency of the hash object lookup routines, you should set the hash table size
according to the amount of data in the hash object.
• The hash table is similar to an array of buckets. If the HASHEXP=4, the hash would have 16 buckets.
This does not limit the hash table to 16 key values. Each bucket can hold an unlimited number of keys.
• When the DATA step adds data to the hash object or retrieves data values from the hash table, the key is
passed to a hash function, which returns the number of the bucket in which to add data or from which
to retrieve data.
• If the number of key and data combinations is larger than the number of buckets, performance might be
reduced because more combinations will be stored per bucket in a binary tree structure.
• If the number of key and data combinations is smaller than the number of buckets, then some of the
buckets will be empty, which wastes memory.
Try different HASHEXP values until you obtain the best result. For example, if the hash object contains a
large number of items, a hash table size of 16 (hashexp=4) is not very efficient. A hash table size of 512
or 1024 (hashexp=9 or 10) results in better performance.

If there is not enough memory in which to load the hash object, the load fails.
Several techniques can be used to determine the amount of memory required by the hash object.
• The size of a hash record is approximately the sum of the sizes of values being placed into the record.
For example, two million 64-byte records take approximately 128 MB. If the SAS system option
MEMSIZE= is set larger than 128 MB and the machine can support executing SAS with at least 128
MB of memory free for loading the hash object, the hash object loads successfully.
• Use the FULLSTIMER SAS system option to determine how much memory the hash object uses with
fewer records. For example, if you load approximately one-third of the records into the hash object,
you can multiply the amount of memory reported by FULLSTIMER by three to determine the
approximate amount of memory needed for the entire hash table. This is an estimate, because the
reported memory usage includes the memory needed to execute the non-hash object portions of the
DATA step.
• The maximum size of the hash object that you can load depends on the maximum amount of memory
addressable per CPU on your particular operating system. For instance, a 4-CPU computer with 8 GB
of memory might limit each CPU to 2 GB of memory. In this case, the maximum size of a hash object
would be less than 2 GB.
Suggestions to avoid memory constraints include the following:
• Subset large data sets before loading the data into the hash object.
• Create a view of a large data set. The view should include syntax that limits the number of columns that
need to be read from the large data set into the hash table.
• Make the length of the hash record as small as possible. For example, instead of the numeric values of
1 and 2, store the values as character '1' and '2'. Numeric data is always stored as 8 bytes in the hash
record.
6-16 Chapter 6 Using DATA Step Hash and Hiter Objects

Declaring a Hash Object

Create a hash object named T.
declare hash T();

Create the T hash object, assign a size, and specify a

return order.
declare hash T(dataset: 'orion.members');

Create the T hash object and load it from

orion.members.
declare hash T(hashexp: 10,
ordered: 'ascending');

The DECLARE statement is an executable statement.

36

In SAS 9.2, you can use data set options in the DECLARE statement when you load the hash object from
a SAS data set.

Example:
declare hash T(dataset: 'orion.members(where=(Code=102
keep=Code Member_Type)';
 6.2 Using Hash Object Methods 6-17

Using Object Dot Syntax

T.definekey('Code');
T.definedata('MemberType');
T.definedone();

General form for object dot method syntax:

OBJECT.METHOD(<arg_tag-1: value-1
<,…arg_tag-n: value-n>>);

object name of the object

method method to invoke
arg_tag name of an argument to be passed
value value of the argument
37

Without the DEFINEDONE method, the log reports the following errors:
ERROR: Method defineDone must be called to complete initialization of hash object before line
189 column 7.
ERROR: DATA STEP Component Object failure. Aborted during the EXECUTION phase.
NOTE: The SAS System stopped processing this step because of errors.

Selected hash object methods available in SAS 9.1 include the following:
DEFINEKEY defines key variables for the hash object.

DEFINEDATA defines data variables for the hash object.

DEFINEDONE completes the initialization of the hash object.

ADD adds key and data values to the hash object.

FIND searches the hash object for a key value, and returns a zero if successful. If the
key is in the hash object, then the FIND method also sets the data variable to
the value of the data item so that it is available for use after the method call.

OUTPUT outputs the hash object’s data values to a SAS data set.

DELETE deletes a hash object.

REPLACE replaces the data for a key in the hash object.

REMOVE removes a key and its associated data from the hash object.

CHECK checks whether the specified key is stored in the hash object.

NUM_ITEMS returns the number of items in the hash object.

6-18 Chapter 6 Using DATA Step Hash and Hiter Objects

Additional hash object methods available in SAS 9.2 include the following:

CLEAR removes all items from the hash object without deleting the hash object instance.

EQUALS determines whether two hash objects are equal.

FIND_NEXT sets the current list item to the next item in the current key's multiple item list and
sets the data for the corresponding data variables.

FIND_PREV sets the current list item to the previous item in the current key's multiple item list
and sets the data for the corresponding data variables.

HAS_NEXT determines whether there is a next item in the current key's multiple data item list.

HAS_PREV determines whether there is a previous item in the current key's multiple data item
list.

REF consolidates the FIND and ADD methods into a single method call.

REMOVEDUP removes the data that is associated with the specified key's current data item from
the hash object.

REPLACEDUP replaces the data that is associated with the current key's current data item with new
data.

SETCUR specifies a starting key item for iteration.

SUM retrieves the summary value for a given key from the hash table and stores the value
in a DATA step variable.

SUMDUP retrieves the summary value for the current data item of the current key and stores
the value in a DATA step variable.
 6.2 Using Hash Object Methods 6-19

Loading Key and Data Values

Use the ADD method to load key and data values into the
hash object.
T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
T.add(key:'30',data:'Internet/Catalog Customers');

38

6.04 Quiz
Why were the statements and methods that instantiate
and load the hash object inside an IF-THEN/DO group?
data mem_type;
length Code $2 MemberType $40;
if _N_=1 then do;
declare hash T();
T.definekey('Code');
T.definedata('MemberType');
T.definedone();
T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
T.add(key:'30',data:'Internet/Catalog Customers');
end;
set orion.europe_customers;
rc1=T.find(key:ThisYrType);
if rc1=0 then ThisYrMember=MemberType;
rc2=T.find(key:LastYrType);
if rc2=0 then LastYrMember=MemberType;
run;

40
6-20 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Cornelia Krahl 20 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . 1

43 ...

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Cornelia Krahl 20 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 . 1

44 ...
 6.2 Using Hash Object Methods 6-21

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Orion Club Gold members Cornelia Krahl 20 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 . 1

45 ...

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Orion Club Gold members Cornelia Krahl 20 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 Orion Club Gold members . 1

46 ...
6-22 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Orion Club Gold members Cornelia Krahl 20 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 Orion Club Gold members 0 Orion Club Gold members 1

50 ...

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold Initialize PDV.
rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Cornelia Krahl 20 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
. . 2

52 ...
 6.2 Using Hash Object Methods 6-23

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Orion Club members Elke Wallstab 10 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 Orion Club Gold members 0 Orion Club members 2

61 ...

Execution
Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke
. . 10 20 T.definekey('Code'); Continue until EOF.
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Orion Club members Ines Deisser 10 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 Orion Club Gold members 0 Orion Club members 10

63 ...
6-24 Chapter 6 Using DATA Step Hash and Hiter Objects

Setup for the Poll

Partial orion.europe_customers
Last This
Customer data mem_type;
. . Yr Yr
_Name
Type Type length Code $2 MemberType $40;
Cornelia
. . 20 20
if _N_=1 then do;
Krahl declare hash T();
Elke T.definekey('Code');
. . 10 20
Wallstab
T.definedata('MemberType');
HASH Object T T.definedone();
KEY: DATA: T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
Code MemberType T.add(key:'30',data:'Internet/Catalog Customers');
Orion Club end;
10 set orion.europe_customers;
members
Orion Club Gold rc1=T.find(key:ThisYrType);
20 if rc1=0 then ThisYrMember=MemberType;
members rc2=T.find(key:LastYrType);
Internet/Catalog if rc2=0 then LastYrMember=MemberType;
30
Customers run;
Partial PDV
Member Customer_ LastYr ThisYr
Code
Type Name ... Type Type
...
Orion Club Gold members Dane Heufmeister 40 20

ThisYr LastYr
rc1 rc2 D _N_
Member Member
0 Orion Club Gold members -2147450842 8
65

6.05 Multiple Choice Poll

What would be the value of LastYrMember if the value of
LastYrType were not found in the hash object (rc2 ne 0)?
a. Missing
b. Orion Club members
c. Orion Club Gold members
d. Internet/Catalog

66
 6.2 Using Hash Object Methods 6-25

6.06 Quiz
Submit the program p306a01 and examine the SAS log.
What are the notes about Code and MemberType?

Why do you get those notes?

69

6.07 Quiz
As the last statement in the DO group in the program
p306a01, add the statement:
call missing(Code, MemberType);

Do the notes disappear?

72
6-26 Chapter 6 Using DATA Step Hash and Hiter Objects

CALL MISSING Routine

The CALL MISSING routine provides initial values to the
variables.

CALL MISSING(varname1<, varname2, ...>);

Argument Type Value Assigned Length

Numeric . 8
Character Blank Current length of
character variable

74

 If the current length of the character variable is any value up to the maximum length, the current
length is not changed. Otherwise, if no length is set for the variable, the current length is set to 1.
 6.2 Using Hash Object Methods 6-27

The FIND Method

The FIND method locates the key value in the hash object
and returns the data values.
General form of the FIND method:

object.FIND(<KEY: keyvalue-1,..., KEY: keyvalue-n>);

 The FIND method returns a numeric value that

indicates whether the FIND method succeeded or
failed.
rc=object.FIND();

75

rc specifies whether the method succeeded or failed.

object specifies the name of the hash object.

KEY: keyvalue specifies the key value whose type must match the corresponding key
variable that is specified in a DEFINEKEY method call. The number of
KEY: keyvalue pairs depends on the number of key variables that you
define by using the DEFINEKEY method.

Retrieving Matching Data

Use the FIND method to retrieve matching data from
the hash object.
rc1=T.find(key:ThisYrType);
if rc1=0 then ThisYrMember=MemberType;
rc2=T.find(key:LastYrType);
if rc2=0 then LastYrMember=MemberType;
Values of the return code variable:
zero success
nonzero failure

If the program does not contain a return code variable for

the method call and the method fails, then an appropriate
error message is written to the log.
76
6-28 Chapter 6 Using DATA Step Hash and Hiter Objects

Exercises

Level 1

1. Using the ADD Method to Create a Hash Object with a Single Key
The following table shows the code that Orion Star uses for each type of order and the description of
the order:

Order Code Sale Type

1 Retail Sale

2 Catalog Sale

3 Internet Sale

The data set orion.orders contains the orders placed.

Partial Listing of orion.orders
Order_ Order_ Delivery_
Obs Order_ID Type Employee_ID Customer_ID Date Date

1 1230058123 1 121039 63 11JAN2003 11JAN2003

2 1230080101 2 99999999 5 15JAN2003 19JAN2003
3 1230106883 2 99999999 45 20JAN2003 22JAN2003
4 1230147441 1 120174 41 28JAN2003 28JAN2003
5 1230315085 1 120134 183 27FEB2003 27FEB2003

a. Write a DATA step that creates a data set named orders.

b. Use the ADD method to create a hash table containing the values of the Order_Type as the key
values and the corresponding Sale_Type as the data values. Use the FIND method to retrieve sale
type based on the variable Order_Type in the data set orion.orders.
c. Keep only the variables Order_ID, Order_Type, and Sale_Type.
d. Print the first five observations from the orders data set.
Partial PROC PRINT Output
Order_
Obs Sale_Type Order_ID Type

1 Retail Sale 1230058123 1

2 Catalog Sale 1230080101 2
3 Catalog Sale 1230106883 2
4 Retail Sale 1230147441 1
5 Retail Sale 1230315085 1
 6.2 Using Hash Object Methods 6-29

Level 2

2. Using the ADD Method with a Composite Key

The following table shows the state code, state name, country code, and country names of the states
and countries for the Orion Star employees:

State State Name Country Country Name

FL Florida US United States

PA Pennsylvania US United States

CA California US United States

AU Australia

a. Write a DATA step to create a data set named emps.

b. Use the ADD method to add the composite key values for State and Country and the data values
of State_Name and Country_Name from the table.
c. Read from the data set orion.employee_addresses.
d. Use the FIND method to perform the table lookup.
Hint: The resulting data set, emps, should have 424 observations. If you do not have 424
observations in the data set, determine why and fix the problem.
e. Print the first 10 observations of the data set emps.
Partial PROC PRINT Output
Partial Data Set emps

Employee_
Obs State_Name Country_Name ID Country

1 Florida United States 121044 US

2 Australia 120145 AU
3 Pennsylvania United States 120761 US
4 California United States 120656 US
5 Pennsylvania United States 121107 US
6 Florida United States 121038 US
7 Florida United States 120273 US
8 California United States 120759 US
9 Florida United States 120798 US
10 Florida United States 121030 US
6-30 Chapter 6 Using DATA Step Hash and Hiter Objects

Level 3

3. Using the ADD Method and Creating a SAS Data Set from a Hash Object
The following table contains the continent ID, the location, and the name of the continent:

Continent ID Continent Name Location

91 North America North
93 Europe North
94 Africa South
95 Asia South
96 Australia/Pacific South

a. Write a DATA step to create a hash object from the values in the table.

 The hash object should be ordered in descending order.

b. After the hash object is created, use the OUTPUT method to create a SAS data set named
continents.
Hint: Consult the SAS OnlineDoc to determine how to use the OUTPUT method.
c. Print the continents data set.
PROC PRINT Output
continents Data Set

Continent_
Obs ID Continent_Name Location

1 96 Australia/Pacific South
2 95 Asia South
3 94 Africa South
4 93 Europe North
5 91 North America North
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-31

6.3 Loading a Hash Object with Data from a SAS Data Set

Objectives
„ Load a hash object from a SAS data set.
„ Use a hash object method to match records.

80

Business Scenario
The data set, orion.supplier, contains demographics
about the suppliers for the products.
Partial Listing of orion.supplier
Sup_
Supplier_ Street_
Supplier_Name Supplier_Address Street_ Country
ID ID
Number
Scandinavian
50 6850100389 Kr. Augusts Gate 13 13 NO
Clothing A/S
109 Petterson AB 8500100286 Blasieholmstorg 1 1 SE
316 Prime Sports Ltd 9250103252 9 Carlisle Place 9 GB
755 Top Sports 3150108266 Jernbanegade 45 45 DK
AllSeasons
772 9260115819 553 Cliffview Dr 553 US
Outdoor Clothing
. . . . . .
. . . . . .
. . . . . .

81
6-32 Chapter 6 Using DATA Step Hash and Hiter Objects

Business Scenario
You need to combine orion.supplier with the data set,
orion.product_list, which contains product information.
Partial Listing of orion.product_list
Supplier_ Product_ Product_
Product_ID Product_Name
ID Level Ref_ID
210000000000 Children . 4 .
210100000000 Children Outdoors . 3 210000000000
Outdoor things,
210100100000 . 2 210100000000
Kids
210200000000 Children Sports . 3 210000000000
210200100000 A-Team, Kids . 2 210200000000
Kids Sweat Round
210200100009 3298 1 210200100000
Neck,Large Logo
. . . . .
. . . . .
. . . The first
five
. values of .
Supplier_ID are missing.

82

Loading Data from a SAS Data Set

data supplier_info;
drop rc;
length Supplier_Name $ 40 Supplier_Address $ 45
Country $ 2;
if _N_=1 then do;
declare hash S(dataset:'orion.supplier');
S.definekey('Supplier_ID');
S.definedata('Supplier_Name',
'Supplier_Address', 'Country');
S.definedone();
call missing(Supplier_Name,
Supplier_Address, Country);
end;
set orion.product_list;
rc=S.find();
if rc=0;
run;
83 p306d02
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-33

Execution
orion.product_list (obs=1)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
210000000000 Children . 4 .

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
. . . 1
84 ...

Execution
orion.product_list (obs=1)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
210000000000 Children . 4 .

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
210000000000 Children . . 1
85 ...
6-34 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
orion.product_list (obs=1)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
210000000000 Children . 4 .

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0; 2147450842
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
210000000000 Children . 2147450842 1
86 ...

Execution
orion.product_list (obs=1)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
210000000000 Children . 4 .

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US
Sports
Ct
rc=S.find();
if rc=0;
False
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
210000000000 Children . 2147450842 1
87 ...
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-35

Execution
orion.product_list (obs=6)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
Kids Sweat Round
210200100009 3298 1 210200100000
Neck,Large Logo

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
S.definekey('Supplier_ID');
Clothing A/S
Gate 13
Blasieh-
Continue until
S.definedata('Supplier_Name',
'Supplier_Address',
109 Petterson AB olmstorg
1
SE
'Country'); _N_=6.
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
210200100009
Kids Sweat Round
Neck,Large Logo 3298 . 6
88 ...

Execution
orion.product_list (obs=6)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
Kids Sweat Round
210200100009 3298 1 210200100000
Neck,Large Logo

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
Kids Sweat Round
210200100009
Neck,Large Logo 3298 0 6
89 ...
6-36 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
orion.product_list (obs=6)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
Kids Sweat Round
210200100009 3298 1 210200100000
Neck,Large Logo

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
Kids Sweat Round
A Team Sports 2687 Julie Ann Ct US 210200100009
Neck,Large Logo 3298 0 6
90 ...

Execution
orion.product_list (obs=6)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
Kids Sweat Round
210200100009 3298 1 210200100000
Neck,Large Logo

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . True
Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
Kids Sweat Round
A Team Sports 2687 Julie Ann Ct US 210200100009
Neck,Large Logo 3298 0 6
91 ...
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-37

Execution
orion.product_list (obs=6)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
Kids Sweat Round
210200100009 3298 1 210200100000
Neck,Large Logo

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
50
Scandinavian
Clothing A/S
Augusts NO Implicit OUTPUT;
S.definekey('Supplier_ID');
Gate 13
Blasieh- Implicit RETURN;
S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
Prime Sports
9 S.definedone();
316 Carlisle GB call missing(Supplier_Name,
Ltd
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
Kids Sweat Round
A Team Sports 2687 Julie Ann Ct US 210200100009
Neck,Large Logo 3298 0 6
92 ...

Execution
orion.product_list (obs=556)
Supplier Product Product_
Product_ID Product_Name
_ID _Level Ref_ID
Top Equipe 99
240800200063 13198 1 210200100000
Black

data supplier_info;
Partial HASH Object S drop rc;
length Supplier_Name $ 40
KEY: DATA: DATA: DATA: Supplier_Address $ 45
Supplier Supplier_ Supplier_ Country
_ID Name Address
Country $ 2;
if _N_=1 then do;
Kr. declare hash S(dataset:'orion.supplier');
Scandinavian
50 Augusts NO
Clothing A/S
Gate 13
S.definekey('Supplier_ID');
Blasieh- S.definedata('Supplier_Name',
109 Petterson AB olmstorg SE 'Supplier_Address',
1 'Country');
S.definedone();
316
Prime Sports
Ltd
9
Carlisle GB call missing(Supplier_Name,Continue until EOF.
Place
. . . .
Supplier_Address,
. . . . Country);
. . . . end;
A Team
2687 set orion.product_list;
3298 Julie Ann US rc=S.find();
Sports
Ct
if rc=0;
run;
Partial PDV
Supplier_ Supplier_ Product Product_ Supplier
rc D_N_
Name Address
Country
_ID Name _ID . . .D
1648 Bloodworth Top Equipe 99
Twain Inc St US 240800200000
Black 13198 0 556
93 ...
6-38 Chapter 6 Using DATA Step Hash and Hiter Objects

Results
proc print data=supplier_info(obs=10);
var Product_ID Supplier_ID Supplier_Name
Supplier_Address Country;
title "Product Information";
run;

Partial PROC PRINT Output

Product Information

Obs Product_ID Supplier_ID Supplier_Name Supplier_Address Country

1 210200100009 3298 A Team Sports 2687 Julie Ann Ct US

2 210200100017 3298 A Team Sports 2687 Julie Ann Ct US
3 210200200022 6153 Nautlius SportsWear Inc 56 Bagwell Ave US
4 210200200023 6153 Nautlius SportsWear Inc 56 Bagwell Ave US
5 210200300006 1303 Eclipse Inc 1218 Carriole Ct US
6 210200300007 1303 Eclipse Inc 1218 Carriole Ct US
7 210200300052 1303 Eclipse Inc 1218 Carriole Ct US
8 210200400020 1303 Eclipse Inc 1218 Carriole Ct US
9 210200400070 1303 Eclipse Inc 1218 Carriole Ct US
10 210200500002 772 AllSeasons Outdoor Clothing 553 Cliffview Dr US

94 p306d02

6.08 Multiple Choice Poll

The program p306d02 created the variable rc and then
dropped it. How can you avoid creating the variable so
that you do not have to drop it?
a. Use a WHERE statement or a WHERE= data set
option.
b. Use a KEEP= or DROP= data set option in
orion.product_list.
c. Test the result of the FIND method in the subsetting
IF statement.
d. Use a KEEP or DROP statement.

96
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-39

Not Creating rc
The program created the variable rc and then dropped it.
How can you avoid creating the variable so that you
do not have to drop it?
data supplier_info;
length Supplier_Name $ 40 Supplier_Address $ 45
Country $ 2;
if _N_=1 then do;
declare hash S(dataset:'orion.supplier');
S.definekey('Supplier_ID');
S.definedata('Supplier_Name',
'Supplier_Address', 'Country');
S.definedone();
call missing(Supplier_Name,
Supplier_Address,
Country);
end;
set orion.product_list;
if S.find()=0;
run;
98 p306d02

6.09 Quiz
How do you know the lengths of the character variables
Supplier_Name, Supplier_Address, and Country?

100
6-40 Chapter 6 Using DATA Step Hash and Hiter Objects

Defining PDV Variables

Instead of the LENGTH statement, you can use an
IF-THEN statement.
data supplier_info;
if _N_=1 then do;
if 0 then set orion.supplier
(keep=Supplier_ID Supplier_Name
Supplier_Address Country);
declare hash S(dataset:'orion.supplier');
S.definekey('Supplier_ID');
S.definedata('Supplier_Name',
'Supplier_Address','Country');
S.definedone();
end;
set orion.product_list;
if S.find()=0;
run;

Because the IF condition is false during execution, the SET

statement is compiled, but not executed. The PDV includes all
the kept variables from orion.supplier. p306d02
102

Using DATA Set Options

In SAS 9.2, you can use SAS DATA set options to limit
the amount of data loaded into a hash object.
data supplier_info;
if _N_=1 then do;
if 0 then set orion.supplier
(keep=Supplier_ID Supplier_Name
Supplier_Address Country);
declare hash S(dataset:"orion.supplier(keep=
Supplier_ID Supplier_Name
Supplier_Address Country
where=(Country='US'))");
S.definekey('Supplier_ID');
S.definedata('Supplier_Name',
'Supplier_Address','Country');
S.definedone();
end;
set orion.product_list;
if S.find()=0;
run;
p306d03
103
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-41

Advantages and Disadvantages of Hash

Objects
Advantages Disadvantages
use of character unique keys required
and numeric keys before SAS 9.2
use of composite keys DATA step only
faster lookup than formats memory requirements
or merges/joins
ability to be loaded
from a SAS data set
fine level of control (flexibility)
ability to do chained lookups

104

Comparing Arrays and Hash Objects

Array Hash Object
The subscript value(s) must be The keys can be character,
numeric. numeric, or both.
One data value can be associated Multiple data items can be
with the subscript value(s). associated with the key value.
An array uses less memory A hash object uses more
than a hash object. memory than an array.
The size of the array is The size of the hash object is
determined at compilation time. determined at execution time.
Subscript values must be The keys do not have to be
consecutive integers. consecutive or sorted.
An array selects values by direct A hash object uses a hash
access based on the subscript function for the lookup process.
value.
Arrays can only be used Hash objects can only be used
105 in the DATA step. in the DATA step.
6-42 Chapter 6 Using DATA Step Hash and Hiter Objects

Exercises

Level 1

4. Loading the Hash Object from a SAS Data Set

The data set orion.customer_type contains the Customer_Type_ID variable and the
Customer_Type variable that is a description of the customer type.
Listing of orion.customer_type
orion.customer_type

Customer_ Customer_
Obs Type_ID Customer_Type Group_ID Customer_Group

1 1010 Orion Club members inactive 10 Orion Club members

2 1020 Orion Club members low activity 10 Orion Club members
3 1030 Orion Club members medium activity 10 Orion Club members
4 1040 Orion Club members high activity 10 Orion Club members
5 2010 Orion Club Gold members low activity 20 Orion Club Gold members
6 2020 Orion Club Gold members medium activity 20 Orion Club Gold members
7 2030 Orion Club Gold members high activity 20 Orion Club Gold members
8 3010 Internet/Catalog Customers 30 Internet/Catalog Customers

The data set orion.customer contains the Customer_ID variable and the Customer_Type_ID
variable.
Partial Listing of orion.customer
Partial orion.customer

Customer_
Obs Customer_ID Type_ID

1 4 1020
2 5 2020
3 9 2020
4 10 1040
5 11 1040
6 12 1030
7 13 2010
8 16 3010
9 17 1030
10 18 1020

a. Write a DATA step to create a data set named customers that reads the variables Customer_ID
and Customer_Type_ID from the data set orion.customer.
b. Create a hash object and load it with the data from orion.customer_type. The key should be the
variable Customer_Type_ID, and the data item should be the variable Customer_Type.
c. Use the hash object to look up the Customer_Type description.
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-43

d. Print the first 10 observations of the customers data set.

Partial PROC PRINT Output
Partial customers Data Set

Customer_
Obs Customer_Type Customer_ID Type_ID

1 Orion Club members low activity 4 1020

2 Orion Club Gold members medium activity 5 2020
3 Orion Club Gold members medium activity 9 2020
4 Orion Club members high activity 10 1040
5 Orion Club members high activity 11 1040
6 Orion Club members medium activity 12 1030
7 Orion Club Gold members low activity 13 2010
8 Internet/Catalog Customers 16 3010
9 Orion Club members medium activity 17 1030
10 Orion Club members low activity 18 1020

Level 2

5. Loading Multiple Hash Objects from SAS Data Sets

The data set orion.product_list contains Product_ID and Product_Name for the products sold.
Partial Listing of orion.product_list
Partial orion.product_list

Obs Product_ID Product_Name

1 210000000000 Children
2 210100000000 Children Outdoors
3 210100100000 Outdoor things, Kids
4 210200000000 Children Sports
5 210200100000 A-Team, Kids

The data set orion.customer_dim contains Customer_ID, Customer_Country, and

Customer_Name for customers who made purchases.
Partial Listing of orion.customer_dim
Partial orion.customer_dim

Customer_
Obs Customer_ID Country Customer_Name

1 4 US James Kvarniq
2 5 US Sandrina Stephano
3 9 DE Cornelia Krahl
4 10 US Karen Ballinger
5 11 DE Elke Wallstab
6-44 Chapter 6 Using DATA Step Hash and Hiter Objects

The data set orion.country contains Country and Country_Name.

Partial Listing of orion.country
Partial orion.country

Country_
Obs Country Name

1 AU Australia
2 CA Canada
3 DE Germany
4 IL Israel
5 TR Turkey

The data set orion.order_fact contains Customer_ID and information about the orders.
Partial Listing of orion.order_fact
Partial orion.order_fact

Order_ Total_Retail_
Obs Customer_ID Date Product_ID Quantity Price

1 63 11JAN2003 220101300017 1 $16.50

2 5 15JAN2003 230100500026 1 $247.50
3 45 20JAN2003 240600100080 1 $28.30
4 41 28JAN2003 240600100010 2 $32.00
5 183 27FEB2003 240200200039 3 $63.60

a. Create a data set named billing that reads Customer_ID, Order_Date, Product_ID, Quantity,
and Total_Retail_Price from orion.order_fact.
b. Create a hash object from orion.product_list with the key Product_ID and the data
Product_Name.
c. Create a hash object from orion.customer_dim with the key Customer_ID and the data
Customer_Country and Customer_Name.
d. Create a hash object from orion.country with the key Country and the data Country_Name.
e. Use the three hash objects to look up Customer_Name, Country_Name, and Product_Name.
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-45

f. Sort the billing data set by Customer_ID and Product_ID and print the first five observations.
Partial PROC PRINT Output
Billing Information
Using a HASH Data Step Object

Customer_
Obs Customer_ID Customer_Name Country Country_Name Product_ID

1 4 James Kvarniq US United States 220101400145

2 4 James Kvarniq US United States 230100100053
3 4 James Kvarniq US United States 240500100017
4 4 James Kvarniq US United States 240500100029
5 4 James Kvarniq US United States 240500200083

Order_ Total_Retail_
Obs Product_Name Date Quantity Price

1 Essence.baseball Cap 16APR2004 1 $16.70

2 Monster Men's Pants with Zipper 18DEC2004 2 $92.60
3 A-team Sweat Round Neck, Small Logo 08APR2004 4 $214.00
4 Men's Sweatshirt w/Hood Big Logo 08APR2004 1 $58.90
5 Force Technical Jacket w/Coolmax 19AUG2004 3 $201.90

Level 3

6. Loading the Hash Object from a SAS Data Set and Retrieving Multiple Values
The data set orion.staff contains the employee ID and the manager ID for that employee.
Partial Listing of orion.staff
Partial orion.staff

Start_
Obs Employee_ID Date End_Date Job_Title Salary

1 120101 01JUL2003 31DEC9999 Director $163,040

2 120102 01JUN1989 31DEC9999 Sales Manager $108,255
3 120103 01JAN1974 31DEC9999 Sales Manager $87,975
4 120104 01JAN1981 31DEC9999 Administration Manager $46,230
5 120105 01MAY1999 31DEC9999 Secretary I $27,110

Birth_ Emp_Hire_ Emp_Term_

Obs Gender Date Date Date Manager_ID

1 M 18AUG1976 01JUL2003 . 120261

2 M 11AUG1969 01JUN1989 . 120101
3 M 22JAN1949 01JAN1974 . 120101
4 F 11MAY1954 01JAN1981 . 120101
5 F 21DEC1974 01MAY1999 . 120101
6-46 Chapter 6 Using DATA Step Hash and Hiter Objects

The data set orion.employee_addresses contains the names of all employees.

Partial Listing of orion.employee_addresses
Partial orion.employee_addresses

Employee_ Street_
Obs ID Employee_Name Street_ID Number

1 121044 Abbott, Ray 9260116912 2267

2 120145 Aisbitt, Sandy 1600101803 30
3 120761 Akinfolarin, Tameaka 9260121030 5
4 120656 Amos, Salley 9260123736 3524
5 121107 Anger, Rose 9260120989 744

Postal_
Obs Street_Name City State Code Country

1 Edwards Mill Rd Miami-Dade FL 33135 US

2 Bingera Street Melbourne 2001 AU
3 Donnybrook Rd Philadelphia PA 19145 US
4 Calico Ct San Diego CA 92116 US
5 Chapwith Rd Philadelphia PA 19142 US

The data set orion.employee_payroll has an employee ID and the salary for each employee.
Partial Listing of orion.employee_payroll
Partial orion.employee_payroll

Employee_ Birth_ Employee_ Employee_ Marital_

Obs Employee_ID Gender Salary Date Hire_Date Term_Date Status Dependents

1 120101 M 163040 6074 15887 . S 0

2 120102 M 108255 3510 10744 . O 2
3 120103 M 87975 -3996 5114 . M 1
4 120104 F 46230 -2061 7671 . M 1
5 120105 F 27110 5468 14365 . S 0

a. Write a DATA step to create a data set named manager that reads the Employee_ID and Salary
variables from orion.employee_payroll.
b. Create hash objects from the data sets orion.employee_addresses and orion.staff.
c. Use the hash object from orion.staff to return the Manager_ID value for each Employee_ID in
orion.employee_payroll.
d. Use the hash object from orion.employee_addresses to retrieve the names for both employees
and the manager for the employees.
 6.3 Loading a Hash Object with Data from a SAS Data Set 6-47

e. Print the first five observations of the manager data set.

Partial PROC PRINT Output
Partial Manager Data Set

Manager_
Obs EmpName ManagerName Employee_ID Salary ID

1 Lu, Patrick Highpoint, Harry 120101 163040 120261

2 Zhou, Tom Lu, Patrick 120102 108255 120101
3 Dawes, Wilson Lu, Patrick 120103 87975 120101
4 Billington, Kareen Lu, Patrick 120104 46230 120101
5 Povey, Liz Lu, Patrick 120105 27110 120101
6-48 Chapter 6 Using DATA Step Hash and Hiter Objects

6.4 Using the DATA Step Hiter Object

Objectives
„ Define a hiter object
„ Investigate the methods for the hiter object
„ Write a DATA step using the hiter object.

109

Defining the Hiter Object

The hiter object is a hash iterator object.
„ Using the hiter object, you can access items in the
hash table based on the location of the value of the
key rather than the value of the key or you can use the
hiter object to move forwards and/or backwards in the
hash object.
„ You cannot define a hiter object without first defining a
hash object.

110
 6.4 Using the DATA Step Hiter Object 6-49

Using the DECLARE Statement for

a Hiter Object
DECLARE HITER iterator-name('hash-name');

declare hash Customer(dataset:'orion.order_fact',

ordered:'descending');
customer.definekey('Total_Retail_Price', 'Customer_ID');
customer.definedata('Total_Retail_Price', 'Customer_ID'
'Product_ID');
customer.definedone();
declare hiter C('customer');

111

Selected Methods for the Hiter Object

FIRST() returns the first data values in the underlying hash
object based on the value of the key value(s).
LAST() returns the last data values in the underlying hash
object based on the value of the key value(s).
NEXT() returns the data values for the next item in key
order in the underlying hash object. A nonzero
value is returned if the next item cannot be
retrieved.
PREV() returns the data values for the previous item in
key order in the underlying hash object. A nonzero
value is returned if the previous item cannot be
retrieved.

112
6-50 Chapter 6 Using DATA Step Hash and Hiter Objects

Business Scenario
The data set orion.order_fact contains the total retail price
of items that were ordered. You need to know the two
customers who ordered the most expensive items and the
two customers who ordered the least expensive items.
Partial Listing of orion.order_fact
Customer Employee Total_Retail CostPrice
Street_ID . . . Discount
_ID _ID _Price _Per_Unit
63 121039 9260125492 . .. $16.50 $7.45 .
5 99999999 9260114570 . .. $247.50 $109.55 .
45 99999999 9260104847 . .. $28.30 $8.55 .
41 120174 1600101527 . .. $32.00 $6.50 .
183 120134 1600100760 . .. $63.60 $8.80 .
. . . . . . .
. . . . . . .
. . . . . . .

113

Hiter Object
data top bottom;
drop i;
if 0 then set orion.order_fact(keep=Customer_ID Product_ID
Total_Retail_Price);
if _N_=1 then do;
declare hash Customer(dataset:'orion.order_fact',
ordered:'descending');
customer.definekey('Total_Retail_Price', 'Customer_ID');
customer.definedata('Total_Retail_Price', 'Customer_ID',
'Product_ID');
customer.definedone();
declare hiter C('customer');
end;
C.first();
do i=1 to 2;
output top;
C.next();
end;
C.last();
do i=1 to 2;
output bottom;
C.prev();
end;
stop;
run;
114 p306d04
 6.4 Using the DATA Step Hiter Object 6-51

Hiter Object
data top bottom;
drop i;
if 0 then set orion.order_fact(keep=Customer_ID Product_ID
Total_Retail_Price);
if _N_=1 then do;
declare hash Customer(dataset:'orion.order_fact',
ordered:'descending');
customer.definekey('Total_Retail_Price', 'Customer_ID');
customer.definedata('Total_Retail_Price', 'Customer_ID',
'Product_ID');
customer.definedone();
declare hiter C('customer');
end;
C.first();
do i=1 to 2;
output top;
C.next();
end;
C.last();
do i=1 to 2;
output bottom;
C.prev();
end;
stop;
run;
115 p306d04

Execution
Partial Hash Object customer
KEY: DATA: data top bottom;
KEY: DATA: DATA: drop i;
Total_ Total_ if 0 then set orion.order_fact
Customer Customer Product_
Retail_ Retail_ (keep=Customer_ID Product_ID
_ID _ID ID Total_Retail_Price);
Price Price
if _N_=1 then do;
16.50 63 16.50 63 220101300017 declare hash
247.50 5 247.50 5 230100500026 customer(dataset:'orion.order_fact',
ordered:'descending');
28.30 45 28.30 45 240600100080 customer.definekey('Total_Retail_Price',
32.00 41 32.00 41 240600100010 'Customer_ID');
. . . . . customer.definedata('Total_Retail_Price',
'Customer_ID',
. . . . . 'Product_ID');
. . . . . customer.definedone();
declare hiter C('customer');
95.10 10 95.10 10 240500200016 end;
48.20 10 48.20 10 240500200122
75.20 89 75.20 89 240700200018
33.80 5 Notice the
33.80 5 220101400130
unordered
hash object.
PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
. . . . 1

116 ...
6-52 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Hiter C View of Partial Hash Object customer
KEY: DATA: data top bottom;
KEY: DATA: DATA: drop i;
Total_ Total_
Customer Customer Product_ if 0 then set orion.order_fact
Retail_ Retail_ (keep=Customer_ID Product_ID
_ID _ID ID Total_Retail_Price);
Price Price
if _N_=1 then do;
1937.20 70100 1937.20 70100 240200100173 declare hash
1796.00 79 1796.00 79 240200100076 customer(dataset:'orion.order_fact',
1687.50 16 1687.50 16 230100700009 ordered:'descending');
customer.definekey('Total_Retail_Price',
1561.80 183 1561.80 183 240300300090 'Customer_ID');
. . . . . customer.definedata('Total_Retail_Price',
'Customer_ID',
. . . . . 'Product_ID');
. . . This is the
. . customer.definedone();
declare hiter C('customer');
3.20 69 3.20 69 230100500004
3.00 5 3.00 hiter object’s
5 240100100433
end;

2.70 11171 2.70 descending

11171 240200100021
2.60 79 2.60 79 230100500045
ordered view
of the hash
object.
PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
. . . . 1

117

Hiter Object
data top bottom;
drop i;
if 0 then set orion.order_fact(keep=Customer_ID Product_ID
Total_Retail_Price);
if _N_=1 then do;
declare hash customer(dataset:'orion.order_fact',
ordered:'descending');
customer.definekey('Total_Retail_Price', 'Customer_ID');
customer.definedata('Total_Retail_Price', 'Customer_ID',
'Product_ID');
customer.definedone();
declare hiter C('customer');
end;
C.first();
do i=1 to 2;
output top;
C.next();
end;
C.last();
do i=1 to 2;
output bottom;
C.prev();
end;
stop;
run;

118 p306d04
 6.4 Using the DATA Step Hiter Object 6-53

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
70100 240200100173 1937.20 . 1

119 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
70100 240200100173 1937.20 1 1

120 ...
6-54 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . C.prev();
. Output. current observation.
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
70100 240200100173 1937.20 1 1

121 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 240200100076 1796.00 1 1

122 ...
 6.4 Using the DATA Step Hiter Object 6-55

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 240200100076 1796.00 2 1

123 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 240200100076 1796.00 2 1

124 ...
6-56 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
c.prev();
.
3.20
.
69
.
3.20
.
69
Output
.
230100500004
current observation.
end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 240200100076 1796.00 2 1

125 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
16 230100700009 1687.50 2 1

126 ...
 6.4 Using the DATA Step Hiter Object 6-57

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
16 230100700009 1687.50 3 1

127 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop; Exit the
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run; DO loop.

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
16 230100700009 1687.50 3 1

128 ...
6-58 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 230100500045 2.60 3 1

129 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 230100500045 2.60 1 1

130 ...
 6.4 Using the DATA Step Hiter Object 6-59

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_ output top;
Retail_ Retail_
_ID _ID ID
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00
2.70
5
11171
3.00
2.70 11171
5 Output
240100100433
240200100021
current observation.
stop;
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
79 230100500045 2.60 1 1

131 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
11171 240200100021 2.70 1 1

132 ...
6-60 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
11171 240200100021 2.70 2 1

133 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
11171 240200100021 2.70 2 1

134 ...
 6.4 Using the DATA Step Hiter Object 6-61

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00
2.70
5
11171
3.00
2.70
Output current observation.
5 240100100433
11171 240200100021
stop;
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
11171 240200100021 2.70 2 1

135 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
5 240100100433 3.00 2 1

136 ...
6-62 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
5 240100100433 3.00 3 1

137 ...

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
. . . . .
. . . . . C.prev();
3.20 69 3.20 69 230100500004 end;
3.00 5 3.00 5 240100100433 stop;
2.70 11171 2.70 11171 240200100021
2.60 79 2.60 79 230100500045
run;

Exit the
DO loop.
PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
5 240100100433 3.00 3 1

138 ...
 6.4 Using the DATA Step Hiter Object 6-63

Execution
Hiter C View of Partial Hash Object customer C.first();
KEY: DATA:
Total_
KEY:
Total_
DATA: DATA: do i=1 to 2;
Customer Customer Product_
Retail_
_ID
Retail_
_ID ID
output top;
Price Price C.next();
1937.20 70100 1937.20 70100 240200100173
1796.00 79 1796.00 79 240200100076
end;
1687.50 16 1687.50 16 230100700009 C.last();
1561.80 183 1561.80 183 240300300090 do i=1 to 2;
. . . . . output bottom;
.The STOP statement prevents
. . . .
. . . . . C.prev();
the following note in the log:
3.20 69 3.20 69 230100500004 end;
NOTE: DATA STEP stopped
3.00
2.70
5
11171
3.00
2.70 11171
5 240100100433
240200100021
stop;
run;
due to looping.
2.60 79 2.60 79 230100500045

PDV
Total_
Product_ D D
Customer_ID Retail_ i _N_
ID
Price
5 240100100433 3.00 3 1

139 ...

p306d04
proc print data=top;
title 'Top 2 Big Spenders';
run;

proc print data=bottom;

title 'Bottom 2 Frugal Spenders';
run;
Output
Top 2 Big Spenders

Total_Retail_
Obs Customer_ID Product_ID Price

1 70100 240200100173 $1,937.20

2 79 240200100076 $1,796.00

Bottom 2 Frugal Spenders

Total_Retail_
Obs Customer_ID Product_ID Price

1 79 230100500045 $2.60
2 11171 240200100021 $2.70
6-64 Chapter 6 Using DATA Step Hash and Hiter Objects

Using the STOP Statement

General form of the STOP statement:

STOP;

„ The STOP statement causes SAS to stop processing

the current DATA step immediately and resume
processing statements after the end of the current
DATA step.
„ SAS outputs a data set for the current DATA step.
„ The observation being processed when the STOP
statement executes is not added.
„ The STOP statement can be used alone or in an
IF-THEN statement or SELECT group.

140
 6.4 Using the DATA Step Hiter Object 6-65

Exercises

Level 1

7. Using a Hiter Object

a. Use the data set orion.shoe_sales that contains the variables Product_ID, Product_Name, and
Total_Retail_Price to create two data sets named expensive and least_expensive. The data set
expensive should contain the five most expensive shoes and the data set least_expensive should
contain the five least expensive shoes.
b. Print each of the data sets.
Listing of expensive
The Five Most Expensive Shoes

Total_Retail_
Obs Product_ID Product_Name Price

1 240200100051 Bretagne Stabilites 2000 Goretex Shoes $420.90

2 220200300129 Torino Men's Leather Adventure Shoes $406.00
3 240200100227 Rubby Women's Golf Shoes w/Gore-Tex $323.80
4 220100700024 Armadillo Road Dmx Women's Running Shoes $313.80
5 240200100225 Rubby Men's Golf Shoes w/Goretex $306.20

Listing of least_expensive
The Five Least Expensive Shoes

Total_Retail_
Obs Product_ID Product_Name Price

1 240100100433 Shoelace White 150 Cm $3.00

2 240100100434 Shoeshine Black $16.40
3 210200400020 Kids Baby Edge Max Shoes $38.00
4 210200400070 Tony's Children's Deschutz (Bg) Shoes $41.60
5 220200100137 Big Guy Men's Multicourt Ii Shoes $50.30
6-66 Chapter 6 Using DATA Step Hash and Hiter Objects

Level 2

8. Using a Hiter Object

a. Use the data set orion.shoe_sales that contains the variables Product_ID, Product_Name, and
Total_Retail_Price to create a data set named shoe_sales that contains the five most expensive
shoes and the five least expensive shoes. The data set should contain a new variable named Rank
that has the value of 'Top 1' to 'Top 5' for the five most expensive and 'Bottom 1' to
'Bottom 5' for the five least expensive shoes.

b. Print the data set.

Listing of shoe_sales
Shoes

Total_Retail_
Obs Product_ID Product_Name Price Rank

1 240200100051 Bretagne Stabilites 2000 Goretex Shoes $420.90 Top 1

2 220200300129 Torino Men's Leather Adventure Shoes $406.00 Top 2
3 240200100227 Rubby Women's Golf Shoes w/Gore-Tex $323.80 Top 3
4 220100700024 Armadillo Road Dmx Women's Running Shoes $313.80 Top 4
5 240200100225 Rubby Men's Golf Shoes w/Goretex $306.20 Top 5
6 240100100433 Shoelace White 150 Cm $3.00 Bottom 1
7 240100100434 Shoeshine Black $16.40 Bottom 2
8 210200400020 Kids Baby Edge Max Shoes $38.00 Bottom 3
9 210200400070 Tony's Children's Deschutz (Bg) Shoes $41.60 Bottom 4
10 220200100137 Big Guy Men's Multicourt Ii Shoes $50.30 Bottom 5

Level 3

9. Using a Hiter Object

a. Use a hiter object to create a data set named different that contains unique values of
Customer_ID and Order_Type from the data set named orion.order_fact. There should be 100
observations in the data set different.
b. Print the first 10 observations of the data set different.
Partial Listing of different
No Duplicates

Order_
Obs Customer_ID Type

1 4 1
2 4 3
3 5 1
4 5 2
5 5 3
6 9 3
7 10 1
8 10 2
9 11 3
10 12 1
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-67

6.5 Using a Hash Object for Chained Lookups (Self-Study)

Objectives
„ Define a chained lookup.
„ Use a hash object to perform a chained lookup.

144

Defining a Chained Lookup

A chained lookup can be one of two types of lookup
operation:
1. using the value of a data variable as the key
to the next FIND method
2. creating a “chain” from more than one variable

145
6-68 Chapter 6 Using DATA Step Hash and Hiter Objects

Business Scenario: Example 1

Because of a computer problem, some customers had to
place an order more than one time. The customer IDs,
along with the product IDs and the dates on which
customers reordered the merchandise, are stored in a
SAS data set named orion.multiple_orders.
Partial Listing of orion.multiple_orders
Customer_ID Product_ID Order_Date
16 220200100035 27AUG2006
16 220200100035 28AUG2006
16 220200100035 30AUG2006
49 210201000126 07APR2007
. . .
. . .
. . .

146

Business Scenario: Example 1

The required data set not only contains the order date,
but also the next order date.
Partial Listing of lookup
Customer_ Next_Order_
Product_ID Order_Date
ID Date
16 220200100035 27AUG2006 28AUG2006
16 220200100035 28AUG2006 30AUG2006
16 220200100035 30AUG2006 .
49 210201000126 07APR2007 08APR2007
49 210201000126 08APR2007 10APR2007
49 210201000126 10APR2007 11APR2007
49 210201000126 11APR2007 .
. . . .
. . . .
. . . .

147
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-69

Business Scenario: Example 1

Partial Listing of orion.multiple_orders Partial Listing of lookup
Customer_ Product_ Order_ Next_
Customer_ Order_
ID ID Date . . . Order_
ID Date
16 220200100035 27AUG2006 Date
16 220200100035 28AUG2006 16 . . . 27AUG2006 28AUG2006
16 220200100035 30AUG2006 16 . . . 28AUG2006 30AUG2006
49 210201000126 07APR2007 16 . . . 30AUG2006 .
49 210201000126 08APR2007 49 . . . 07APR2007 08APR2007
49 210201000126 10APR2007 49 . . . 08APR2007 10APR2007
49 210201000126 11APR2007 49 . . . 10APR2007 11APR2007
70108 240200200071 22JUL2007 49 . . . 11APR2007 .
70108 240200200071 25JUL2007 70108 22JUL2007 25JUL2007
70108 240200200071 26JUL2007 70108 25JUL2007 26JUL2007
70108 240200200071 28JUL2007 70108 26JUL2007 28JUL2007
70108 240200200071 30JUL2007 70108 28JUL2007 30JUL2007
70108 240200200071 01AUG2007 70108 30JUL2007 01AUG2007
70108 240200200071 02AUG2007 70108 01AUG2007 02AUG2007
70108 240200200071 05AUG2007 70108 02AUG2007 05AUG2007
70165 240200100050 08SEP2007 70108 05AUG2007 .
. . . . . . .
. . . . . . .
. . . . . . .

148

Example 1
proc sort data=orion.multiple_orders
out=multiple_orders;
by Customer_ID;
run;

data multiple_orders;
set multiple_orders;
rename Order_Date=OD;
ObsNum=_N_;
run;

p306d05
149 continued...
6-70 Chapter 6 Using DATA Step Hash and Hiter Objects

Example 1
data lookup;
format Next_Order_Date date9.;
keep Customer_ID Product_ID Order_Date
Next_Order_Date;
if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
LU.definekey('ObsNum');
LU.definedata('OD');
LU.definedone();
call missing(OD);
end;
set multiple_orders(rename=(OD=Order_Date));
by Customer_ID;
Obs=ObsNum + 1;
rc=LU.find(key:Obs);
if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
run;
p306d05
150

6.10 Quiz
What is the purpose of the BY statement in the DATA
step?

152
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-71

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. . . . . .

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 1 . . 1
154 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
multiple_orders.sas7bdat
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. . 16 220200100035 27AUG2006 1

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 0 . . 1
155 ...
6-72 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. . 16 220200100035 27AUG2006 1

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 0 2 . 1
156 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. 28AUG2006 16 220200100035 27AUG2006 1

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 0 2 0 1
157 ...
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-73

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
28AUG2006 28AUG2006 16 220200100035 27AUG2006 1

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 0 2 0 1
158 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
False
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
28AUG2006 28AUG2006 16 220200100035 27AUG2006 1

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 0 2 0 1
159 ...
6-74 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone(); Implicit OUTPUT;
call missing(OD);
4 07APR2007 end; Implicit RETURN;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
28AUG2006 28AUG2006 16 220200100035 27AUG2006 1

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
1 0 2 0 1
160 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. . 16 220200100035 28AUG2006 2

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 0 3 . 2
161 ...
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-75

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. 30AUG2006 16 220200100035 28AUG2006 2

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 0 3 0 2
162 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
30AUG2006 30AUG2006 16 220200100035 28AUG2006 2

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 0 3 0 2
163 ...
6-76 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
False
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
30AUG2006 30AUG2006 16 220200100035 28AUG2006 2

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 0 3 0 2
164 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone(); Implicit OUTPUT;
call missing(OD); Implicit RETURN;
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
30AUG2006 30AUG2006 16 220200100035 28AUG2006 2

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 0 3 0 2
165 ...
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-77

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. . 16 220200100035 30AUG2006 3

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 4 . 3
166 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. 07APR2007 16 220200100035 30AUG2006 3

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 4 0 3
167 ...
6-78 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
07APR2007 07APR2007 16 220200100035 30AUG2006 3

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 4 0 3
168 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end; True
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
07APR2007 07APR2007 16 220200100035 30AUG2006 3

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 4 0 3
169 ...
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-79

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. 07APR2007 16 220200100035 30AUG2006 3

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 4 0 3
170 ...

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD'); Implicit OUTPUT;
3 30AUG2006 LU.definedone();
call missing(OD); Implicit RETURN;
4 07APR2007 end;
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. 07APR2007 16 220200100035 30AUG2006 3

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 4 0 3
171 ...
6-80 Chapter 6 Using DATA Step Hash and Hiter Objects

Execution
data lookup;
Partial Hash Object LU format Next_Order_Date date9.;
ObsNum OD keep Customer_ID Product_ID Order_Date
Next_Order_Date;
1 27AUG2006 if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
2 28AUG2006 LU.definekey('ObsNum');
LU.definedata('OD');
3 30AUG2006 LU.definedone();
call missing(OD);
4 07APR2007 end; Continue until EOF.
set multiple_orders(rename=(OD=Order_Date));
5 08APR2007 by Customer_ID;
. . Obs=ObsNum+1;
rc=LU.find(key:Obs);
. . if rc=0 then Next_Order_Date=OD;
if last.Customer_ID then Next_Order_Date=.;
. . run;

PDV
Next_
Customer Product_ Order_
Order_ OD D ObsNum
_ ID ID Date
Date
...
. . 70165 240200100050 19SEP2007 32

First. Last.
D D D Obs D rc D _N_
Customer_ID Customer_ID
0 1 33 -2147450842 32
172

Chained Lookup
proc print data=lookup(obs=10);
var Customer_ID Order_Date Next_Order_Date;
title 'Chained Lookup Example';
run;
PROC PRINT Output
Chained Lookup Example
Next_
Customer_ Order_ Order_
Obs ID Date Date

1 16 27AUG2006 28AUG2006
2 16 28AUG2006 30AUG2006
3 16 30AUG2006 .
4 49 07APR2007 08APR2007
5 49 08APR2007 10APR2007
6 49 10APR2007 11APR2007
7 49 11APR2007 .
8 79 27SEP2007 30SEP2007
9 79 30SEP2007 01OCT2007
10 79 01OCT2007 . p306d05
173
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-81

Business Scenario: Example 2

Suppose that you need to create a list of the Order_Date
values, that is, the date on which a customer placed an
order for a product.

Partial orion.multiple_orders
Customer_ID Product_ID Order_Date
16 220200100035 27AUG2006
16 220200100035 28AUG2006
16 220200100035 30AUG2006
49 210201000126 07APR2007
. . .
. . .
. . .

174

Business Scenario: Example 2

The new data set contains a list of all the dates on which
the orders were placed.

Listing of lookup
Customer_
All_Dates Product_ID
ID
27AUG2006, 28AUG2006, 30AUG2006 16 220200100035
07APR2007, 08APR2007, 10APR2007, 11APR2007 49 210201000126
27SEP2007, 30SEP2007, 01OCT2007 79 240500100057
31AUG2007, 05SEP2007, 08SEP2007,
171 230100500004
10SEP2007, 11SEP2007, 13SEP2007, 14SEP2007
29JAN2007, 01FEB2007 2806 240100400058
22JUL2007, 25JUL2007, 26JUL2007,
28JUL2007, 30JUL2007, 01AUG2007, 70108 240200200071
02AUG2007, 05AUG2007
08SEP2007, 10SEP2007, 16SEP2007,
70165 240200100050
18SEP2007, 19SEP2007

175
6-82 Chapter 6 Using DATA Step Hash and Hiter Objects

Creating a List of Values

p306d06
proc sort data=orion.multiple_orders out=multiple_orders;
by Customer_ID;
run;

data multiple_orders;
set multiple_orders;
rename Order_Date=OD;
ObsNum=_N_;
run;

data lookup;
length All_Dates $200;
keep Customer_ID Product_ID All_Dates;
if _N_=1 then do;
declare hash LU(dataset: "multiple_orders");
LU.definekey('ObsNum', 'Customer_ID');
LU.definedata('OD');
LU.definedone();
call missing(OD);
end;
do until (Last);
set multiple_orders(rename=(OD=Order_Date)) end=Last;
by Customer_ID;
if first.Customer_ID then All_Dates=put(Order_Date, date9.);
Obs=ObsNum + 1;
rc=LU.find(key:Obs, key:Customer_ID);
if rc=0 then
All_Dates=catx(', ', All_Dates, put(OD, date9.));
else output;
end;
run;

proc print data=lookup;

var Customer_ID All_Dates;
title 'Chained Lookup Example';
run;
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-83

PROC PRINT Output

Chained Lookup Example

Customer_
Obs ID

1 16
2 49
3 79
4 171
5 2806
6 70108
7 70165

Obs All_Dates

1 27AUG2006, 28AUG2006, 30AUG2006

2 07APR2007, 08APR2007, 10APR2007, 11APR2007
3 27SEP2007, 30SEP2007, 01OCT2007
4 31AUG2007, 05SEP2007, 08SEP2007, 10SEP2007, 11SEP2007, 13SEP2007, 14SEP2007
5 29JAN2007, 01FEB2007
6 22JUL2007, 25JUL2007, 26JUL2007, 28JUL2007, 30JUL2007, 01AUG2007, 02AUG2007, 05AUG2007
7 08SEP2007, 10SEP2007, 16SEP2007, 18SEP2007, 19SEP2007

This problem can also be solved using FIRST. and LAST. processing in the DATA step.
p306d07
proc sort data=orion.multiple_orders out=multiple_orders;
by Customer_ID;
run;

data lookup;
retain All_Dates;
length All_Dates $200;
keep Customer_ID Product_ID All_Dates;
set multiple_orders;
by Customer_ID;
if first.Customer_ID then All_Dates=put(Order_Date, date9.);
Next_Date=lag(Order_Date);
if not first.Customer_ID then All_Dates=
catx(', ', All_Dates, put(Next_Date, date9.));
if last.Customer_ID then output;
run;

proc print data=lookup;

title 'First. Last. Lookup Example';
run;

title;
6-84 Chapter 6 Using DATA Step Hash and Hiter Objects

Exercises

Level 1

10. Using a Hash Object to Locate the Next Value of a Variable

a. Use the data set orion.order_fact that contains the variables Customer_ID, Product_ID, and
Total_Retail_Price to create a data set named next_products, which contains two new variables,
Next_Product_ID and Next_Price. Next_Product_ID is the product that the customer ordered
after the current product. Next_Price is the price of that product.
b. Print the first 10 observations of next_products. Format Next_Price with dollar signs and two
digits to the right of the decimal point.
Next Product Ordered

Total_Retail_ Next_ Next_

Obs Customer_ID Product_ID Price Product_ID Price

1 4 240800200030 $47.70 240600100017 $53.00

2 4 240600100017 $53.00 240700200019 $16.90
3 4 240700200019 $16.90 240500100017 $214.00
4 4 240500100017 $214.00 240500100029 $58.90
5 4 240500100029 $58.90 220101400145 $16.70
6 4 220101400145 $16.70 240700100011 $80.97
7 4 240700100011 $80.97 240500200083 $201.90
8 4 240500200083 $201.90 230100100053 $92.60
9 4 230100100053 $92.60 . .
10 5 230100500026 $247.50 240100100433 $3.00
 6.5 Using a Hash Object for Chained Lookups (Self-Study) 6-85

Level 2

11. Using a Hash Object to Create a Chain of Values

a. Use the data set orion.customer that contains the variables Country and Customer_ID to create
a data set named customer_list that contains a new variable, All_Customers, which is a list of
customers for each country.
b. Open and submit the program p306e11 that contains a PROC REPORT step.
p306e11
proc report data=customer_list nowd headline headskip;
column Country All_Customers;
define Country / width=20 order 'Customer/Country';
define All_Customers / width=50 flow 'Customer/List';
break after Country / skip;
run;
Listing of customer_list
Next Product Ordered

Customer Customer
Country List
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

AU 29, 41, 53, 111, 171, 183, 195, 215

CA 11171, 17023, 26148, 46966, 54655, 70046, 70059,

70079, 70100, 70108, 70165, 70187, 70201, 70210,
70221

DE 9, 11, 13, 16, 19, 33, 42, 50, 61, 65

IL 12386, 14104, 14703, 19444, 19873

TR 544, 908, 928, 1033, 1100, 1684, 2788

US 4, 5, 10, 12, 17, 18, 20, 23, 24, 27, 31, 34, 36,
39, 45, 49, 52, 56, 60, 63, 69, 71, 75, 79, 88,
89, 90, 92

ZA 2550, 2618, 2806, 3959

Level 3

12. Using a Hash Object to Create a Chain of Values

a. Use the data set orion.product_dim to create a data set named suppliers that contains two
variables, All_Products and All_Names. All_Products is a list of all the Product_ID values for
each supplier. All_Names is a list of the names of all of those products. Ensure that none of the
values is truncated.
6-86 Chapter 6 Using DATA Step Hash and Hiter Objects

b. Write a PROC REPORT step to display the data set suppliers. Ensure that the entire lists for
All_Products and All_Names are printed.
Partial Listing of suppliers
Supplier Product List

Product Names of
Supplier List Products
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

3Top Sports 210201000050, 210201000067, Kid Children's T-Shirt, Logo

210201000126, 210201000198, Coord.Children's Sweatshirt,
210201000199, 220101400004, Toddler Footwear Socks with
220101400017, 220101400018, Knobs, South Peak Junior
220101400047, 220101400060, Training Shoes, Starlite Baby
220101400061, 220101400088, Shoes, Badminton Cotton,
220101400091, 220101400092, Men's Cap, Men's Running Tee
220101400098, 220101400117, Short Sleeves, Swimming
220101400130, 220101400138, Trunks Struc, 2bwet 3 Cb
220101400145, 220101400148, Swimming Trunks, 2bwet 3
220101400150, 220101400152, Solid Bikini, Casual Genuine
220101400201, 220101400216, Polo-Shirt, Casual Genuine
220101400237, 220101400238, Tee, Casual Logo Men's
220101400265, 220101400269, Sweatshirt, Casual Sport
220101400276, 220101400285, Shorts, Casual.st.polo
220101400289, 220101400290, Long-sleeved Polo-shirt,
220101400306, 220101400310, Comp. Women's Sleeveless
220101400328, 2201014003 Polo, Dima 2-Layer Men's
Suit, Essence.baseball Cap,
Essence.cap Men's Bag,
Essential Suit 2 Swim Suit,
Essential Trunk 2 Swimming
Trunks, Kaitum Women's Swim
Suit, Mm Daypouch Shoulder
Bag, Mns.jacket Jacket,
Mns.long Tights, Ottis Pes
Men's Pants, Outfit Women's
Shirt, Pine Sweat with Hood,
Quali Jacket with Hood, Quali
Sweatpant, Quali Sweatshirt,
Sherpa Pes Shiny Cotton,
Short Women's Tights, Stars
Swim Suit, Tims Shorts,
Tracker Fitness Stockings,
Pytossage Bathing Sandal,
Liga Football Boot, Men's
Running Shoes Piedmmont,
Hilly Women's Crosstrainer
Shoes, Indoor Handbold
Special Shoes, Mns.raptor
Precision Sg Football, South
Peak Men's Running Shoes,
Torino Men's Leather
Adventure Shoes, T-Shirt
 6.6 Chapter Review 6-87

6.6 Chapter Review

Chapter Review
1. Describe a hash object.

2. When is a hash object deleted from memory?

3. What are the two types of DATA step component

objects?

4. What is a key component?

5. What is the purpose of the DECLARE statement?

179

Chapter Review
6. What is the purpose of the FIND method?

7. What value does the FIND method return when it

executes?

8. Are the key(s) and data item(s) variables in the PDV?

181
6-88 Chapter 6 Using DATA Step Hash and Hiter Objects

Chapter Review
9. Why are the DECLARE, DEFINEKEY, DEFINEDATA,
and DEFINEDONE methods executed in the IF _N_=1
THEN/DO group?

10. Is the DEFINEDONE method required?

183
 6.7 Solutions 6-89

6.7 Solutions

Solutions to Exercises
1. Using the ADD Method to Create a Hash Object with a Single Key
a. Write a DATA step that creates a data set named orders.
b. Use the ADD method to create a hash table containing the values of the Order_Type as the key
values and the corresponding Sale_Type as the data values. Use the FIND method to retrieve sale
type based on the variable Order_Type in the data set orion.orders.
c. Keep only the variables Order_ID, Order_Type, and Sale_Type.
d. Print the first five observations from the orders data set.
p306s01
data orders;
length Sale_Type $40;
keep Order_ID Order_Type Sale_Type;
if _N_=1 then do;
declare hash Product();
Product.definekey('Order_Type');
Product.definedata('Sale_Type');
Product.definedone();
Product.add(key:1, data:'Retail Sale');
Product.add(key:2, data:'Catalog Sale');
Product.add(key:3, data:'Internet Sale');
call missing(Sale_Type);
end;
set orion.orders;
rc=Product.find();
if rc=0;
run;

proc print data=orders(obs=5);

title;
run;
2. Using the ADD Method with a Composite Key
a. Write a DATA step to create a data set named emps.
b. Use the ADD method to add the composite key values for State and Country and the data values
of State_Name and Country_Name from the table.
c. Read from the data set orion.employee_addresses.
d. Use the FIND method to perform the table lookup.
6-90 Chapter 6 Using DATA Step Hash and Hiter Objects

e. Print the first 10 observations of the data set emps.

p306s02
data emps;
length State_Name $ 12 Country_Name $30;
keep Employee_ID Country Country_Name State_Name;
if _N_=1 then do;
declare hash C();
C.definekey('State', 'Country');
C.definedata('State_Name', 'Country_Name');
C.definedone();
C.add(key:'FL',key:'US',data:'Florida',data:'United States');
C.add(key:'PA',key:'US',data:'Pennsylvania',data:'United States');
C.add(key:'CA',key:'US',data:'California',data:'United States');
C.add(key:' ', key:'AU',data:' ',data:'Australia');
call missing(State_Name, Country_Name);
end;
set orion.employee_addresses;
rc=C.find(key:upcase(State), key:upcase(Country));
if rc=0;
run;
proc print data=emps(obs=10);
title;
run;
3. Using the ADD Method and Creating a SAS Data Set from a Hash Object
a. Write a DATA step to create a hash object from the values in the table.
b. After the hash object is created, use the OUTPUT method to create a SAS data set named
continents.
c. Print the continents data set.
p306s03
data _null_;
length Continent_Name $40 Location $5 Continent_ID 8;
if _N_=1 then do;
declare hash C(ordered:'descending');
C.definekey('Continent_ID');
C.definedata('Continent_ID', 'Continent_Name', 'Location');
C.definedone();
C.add(key:91,data:91,data:'North America',data:'North');
C.add(key:93,data:93,data:'Europe',data:'North');
C.add(key:94,data:94,data:'Africa',data:'South');
C.add(key:95,data:95,data:'Asia', data:'South');
C.add(key:96,data:96,data:'Australia/Pacific',
data:'South');
call missing(Continent_ID, Continent_Name, Location);
end;
C.output(dataset:"continents");
run;
(Continued on the next page.)
 6.7 Solutions 6-91

proc print data=continents;

title 'continents Data Set';
run;

title;
4. Loading the Hash Object from a SAS Data Set
a. Write a DATA step to create a data set named customers that reads the variables Customer_ID
and Customer_Type_ID from the data set orion.customer.
b. Create a hash object and load it with the data from orion.customer_type. The key should be the
variable Customer_Type_ID, and the data item should be the variable Customer_Type.
c. Use the hash object to look up the Customer_Type description.
d. Print the first 10 observations of the customers data set.
p306s04
data customers;
length Customer_Type $40;
keep Customer_ID Customer_Type_ID Customer_Type;
if _N_=1 then do;
declare hash Customer(dataset:'orion.customer_type');
Customer.definekey('Customer_Type_ID');
Customer.definedata('Customer_Type');
Customer.definedone();
call missing(Customer_Type);
end;
set orion.customer;
if Customer.find()=0;
run;

proc print data=customers(obs=10);

title 'customers';
run;

/* alternate solution */
data customers;
keep Customer_ID Customer_Type_ID Customer_Type;
if 0 then set orion.customer_type(keep=Customer_Type_ID
Customer_Type);
if _N_=1 then do;
declare hash Customer(dataset:'orion.customer_type');
Customer.definekey('Customer_Type_ID');
Customer.definedata('Customer_Type');
Customer.definedone();
call missing(Customer_Type);
end;
set orion.customer;
if Customer.find()=0;
run;
6-92 Chapter 6 Using DATA Step Hash and Hiter Objects

5. Loading Multiple Hash Objects from SAS Data Sets

a. Create a data set named billing that reads Customer_ID, Order_Date, Product_ID, Quantity,
and Total_Retail_Price from orion.order_fact.
b. Create a hash object from orion.product_list with the key Product_ID and the data
Product_Name.
c. Create a hash object from orion.customer_dim with the key Customer_ID and the data
Customer_Country and Customer_Name.
d. Create a hash object from orion.country with the key Country and the data Country_Name.
e. Use the three hash objects to look up Customer_Name, Country_Name, and Product_Name.
f. Sort the billing data set by Customer_ID and Product_ID and print the first five observations.
p306s05
data billing;
drop rc1 rc2 rc3 Country;
if _N_=1 then do;
if 0 then set orion.product_list(keep=Product_ID
Product_Name);
if 0 then set orion.customer_dim(keep=Customer_ID
Customer_Country
Customer_Name);
if 0 then set orion.country(keep=Country Country_Name);
declare hash Prod(dataset:'orion.product_list');
Prod.definekey('Product_ID');
Prod.definedata('Product_Name');
Prod.definedone();
declare hash Customer(dataset:'orion.customer_dim');
Customer.definekey('Customer_ID');
Customer.definedata('Customer_Country', 'Customer_Name');
Customer.definedone();
declare hash C(dataset:'orion.country');
C.definekey('Country');
C.definedata('Country_Name');
C.definedone();
end;
set orion.order_fact(keep=Order_Date Quantity Product_ID
Total_Retail_Price Customer_ID);
rc1=Customer.find();
if rc1=0;
rc2=Prod.find();
if rc2=0;
rc3=C.find(key:Customer_Country);
if rc3=0;
run;
(Continued on the next page.)
 6.7 Solutions 6-93

proc sort data=billing;

by Customer_ID Product_ID;
run;

proc print data=billing(obs=5);

var Customer_ID Customer_Name Customer_Country
Country_Name Product_ID Product_Name Order_Date
Quantity Total_Retail_Price;
title1 'Billing Information';
title2 'Using a HASH Data Step Object';
run;
6. Loading the Hash Object from a SAS Data Set and Retrieving Multiple Values
a. Write a DATA step to create a data set named manager that reads the Employee_ID and Salary
variables from orion.employee_payroll.
b. Create hash objects from the data sets orion.employee_addresses and orion.staff.
c. Use the hash object from orion.staff to return the Manager_ID for each Employee_ID in
orion.employee_payroll.
d. Use the hash object from orion.employee_addresses to retrieve the names for both employees
and the manager for the employees.
e. Print the first five observations of the manager data set.
p306s06
data manager;
length Employee_Name EmpName ManagerName $40;
keep Employee_ID EmpName Manager_ID ManagerName Salary;
if _N_=1 then do;
declare hash M(dataset:'orion.staff');
M.definekey('Employee_ID');
M.definedata('Manager_ID');
M.definedone();
declare hash N(dataset:'orion.employee_addresses');
N.definekey('Employee_ID');
N.definedata('Employee_Name');
N.definedone();
call missing(Employee_Name);
end;
set orion.employee_payroll(keep=Employee_ID Salary);
rc1=M.find(key:Employee_ID);
rc2=N.find(key:Employee_ID);
if rc2=0 then EmpName=Employee_Name;
else EmpName=' ';
rc3=N.find(key:Manager_ID);
if rc3=0 then ManagerName=Employee_Name;
else ManagerName=' ';
run;
(Continued on the next page.)
6-94 Chapter 6 Using DATA Step Hash and Hiter Objects

proc print data=manager(obs=5);

title "Manager Data Set";
run;
7. Using a Hiter Object
a. Use the data set orion.shoe_sales that contains the variables Product_ID, Product_Name, and
Total_Retail_Price to create two data sets named expensive and least_expensive. The data set
expensive should contain the five most expensive shoes and the data set least_expensive should
contain the five least expensive shoes.
p306s07
data expensive least_expensive;
drop i;
if 0 then set orion.shoe_sales;
if _N_=1 then do;
declare hash Shoes(dataset:'orion.shoe_sales',
ordered:'descending');
Shoes.definekey('Total_Retail_Price');
Shoes.definedata('Total_Retail_Price',
'Product_ID', 'Product_Name');
Shoes.definedone();
declare hiter S('Shoes');
end;

S.first();
do i=1 to 5;
output expensive;
S.next();
end;

S.last();
do i=1 to 5;
output least_expensive;
S.prev();
end;
stop;
run;
b. Print each of the data sets.
p306s07
proc print data=expensive;
title "The Five Most Expensive Shoes";
run;

proc print data=least_expensive;

title "The Five Least Expensive Shoes";
run;
 6.7 Solutions 6-95

8. Using a Hiter Object

a. Use the data set orion.shoe_sales that contains the variables Product_ID, Product_Name, and
Total_Retail_Price to create a data set named shoe_sales that contains the five most expensive
shoes and the five least expensive shoes. The data set should contain a new variable named Rank
that has the value of 'Top 1' to 'Top 5' for the five most expensive and 'Bottom 1' to
'Bottom 5' for the five least expensive shoes.

b. Print the data set.

p306s08
data shoe_sales;
drop i;
if 0 then set orion.shoe_sales;
length Rank $ 8;
if _N_=1 then do;
declare hash Shoes(dataset:'orion.shoe_sales',
ordered:'descending');
Shoes.definekey('Total_Retail_Price');
Shoes.definedata('Total_Retail_Price',
'Product_ID', 'Product_Name');
Shoes.definedone();
declare hiter S('Shoes');
end;

S.first();
do i=1 to 5;
Rank=catx(' ', 'Top', i);
output;
S.next();
end;

S.last();
do i=1 to 5;
Rank=catx(' ', 'Bottom', i);
output;
S.prev();
end;
stop;
run;

proc print data=shoe_sales;

title "Shoes";
run;
6-96 Chapter 6 Using DATA Step Hash and Hiter Objects

9. Using a Hiter Object

a. Use a hiter object to create a data set named different that contains unique values of
Customer_ID and Order_Type from the data set named orion.order_fact. There should be 100
observations in the data set different.
b. Print the first 10 observations of the data set different.
p306s09
data different;
drop rc;
if _N_=1 then do;
if 0 then set orion.order_fact(keep=Customer_ID
Order_Type);
declare hash Orders(dataset: 'orion.order_fact',
ordered: 'yes');
declare hiter O_F('Orders');
orders.defineKey('Customer_ID', 'Order_Type');
orders.defineData('Customer_ID', 'Order_Type');
orders.defineDone();
end;
rc=O_F.first();
do while (rc=0);
output;
rc=O_F.next();
end;
stop;
run;

proc print data=different(obs=10);

title "No Duplicates";
run;
 6.7 Solutions 6-97

10. Using a Hash Object to Locate the Next Value of a Variable

a. Use the data set orion.order_fact that contains the variables Customer_ID, Product_ID, and
Total_Retail_Price to create a data set named next_products, which contains two new variables,
Next_Product_ID and Next_Price. Next_Product_ID is the product that the customer ordered
after the current product. Next_Price is the price of that product.
b. Print the first 10 observations of next_products.
p306s10
proc sort data=orion.order_fact(keep=Customer_ID Product_ID
Total_Retail_Price)
out=order_fact;
by Customer_ID;
run;

data order_fact;
set order_fact;
rename Product_ID=PID Total_Retail_Price=TRP;
ObsNum=_N_;
run;

data next_products;
keep Customer_ID Product_ID Total_Retail_Price
Next_Product_ID Next_Price;
if _N_=1 then do;
declare hash Lu(dataset: "order_fact");
Lu.definekey('ObsNum');
Lu.definedata('PID', 'TRP');
Lu.definedone();
call missing(PID, TRP);
end;
set order_fact(rename=(PID=Product_ID
TRP=Total_Retail_Price));
by Customer_ID;
Obs=ObsNum + 1;
rc=Lu.find(key:Obs);
if rc=0 then do;
Next_Product_ID=PID;
Next_Price=TRP;
end;
if last.Customer_ID then do;
Next_Product_ID=.;
Next_Price=.;
end;
run;

proc print data=next_products(obs=10);

title 'Next Product Ordered';
format Next_Price dollar8.2;
run;
6-98 Chapter 6 Using DATA Step Hash and Hiter Objects

11. Using a Hash Object to Create a Chain of Values

a. Use the data set orion.customer that contains the variables Country and Customer_ID to create
a data set named customer_list that contains a new variable, All_Customers, which is a list of
customers for each country.
p306s11
proc sort data=orion.customer(keep=Country Customer_ID)
out=customers;
by Country;
run;

data customers;
set customers;
ObsNum=_N_;
run;

data customer_list;
length All_Customers $500;
if _N_=1 then do;
declare hash Lu(dataset: "customers");
Lu.definekey('ObsNum','Country');
Lu.definedata('Country','Customer_ID');
Lu.definedone();
end;
do until (Last);
set customers end=Last;
by Country;
if first.Country then All_Customers=Customer_ID;
Obs=ObsNum + 1;
rc=Lu.find(key:Obs, key:Country);
if rc=0 then
All_Customers=catx(', ', All_Customers, Customer_ID);
else output;
end;
run;
b. Open and submit the program p306e11 that contains a PROC REPORT step.
p306e11
proc report data=customer_list nowd headline headskip;
column Country All_Customers;
define Country / width=20 order 'Customer/Country';
define All_Customers / width=50 flow 'Customer/List';
break after Country / skip;
run;
 6.7 Solutions 6-99

12. Using a Hash Object to Create a Chain of Values

a. Use the data set orion.product_dim to create a data set named suppliers that contains two
variables, All_Products and All_Names. All_Products is a list of all the Product_ID values for
each supplier. All_Names is a list of the names of all of those products. Ensure that none of the
values is truncated.
p306s12
proc sort data=orion.product_dim out=product_dim;
by Supplier_ID;
run;

data product_dim;
set product_dim;
ObsNum=_N_;
run;

data suppliers;
length All_Products $500 All_Names $750;
if _N_=1 then do;
declare hash Lu(dataset: "product_dim");
Lu.definekey('ObsNum', 'Supplier_ID');
Lu.definedata('Supplier_Name','Product_ID','Product_Name');
Lu.definedone();
end;
do until (Last);
set product_dim end=Last;
by Supplier_ID;
if first.Supplier_ID then do;
All_Products=Product_ID;
All_Names=Product_Name;
end;
Obs=ObsNum + 1;
rc=Lu.find(key:Obs, key:Supplier_ID);
if rc=0 then do;
All_Products=catx(', ', All_Products, Product_ID);
All_Names=catx(', ', All_Names, Product_Name);
end;
else output;
end;
run;
b. Write a PROC REPORT step to display the data set suppliers. Ensure that the entire lists for
All_Products and All_Names are printed.
proc report data=suppliers nowd headline headskip ls=132;
column Supplier_Name All_Products All_Names;
define Supplier_Name / width=30 order 'Supplier';
define All_Products / width=30 flow 'Product/List';
define All_Names / width=50 flow 'Names of/Products';
break after Supplier_Name / skip;
run;
6-100 Chapter 6 Using DATA Step Hash and Hiter Objects

Solutions to Student Activities (Polls/Quizzes)

6.02 Multiple Answer Poll – Correct Answers

Which of the following could be used to assign
descriptions to each of the variables ThisYrType
and LastYrType?
a. Merging
b. Formats
c. IF-THEN/ELSE
d. Arrays

Merging cannot be used because the descriptions for

Member Type are not stored in a SAS data set.
Arrays cannot be used because the values of the
variables ThisYrType and LastYrType are not
consecutive integers.
17

6.03 Multiple Answer Poll – Correct Answers

When would you not use quotation marks around the
value for an argument to a method?
a. The value is numeric.
b. The value is character.
c. The value of a PDV variable is wanted.
d. The value is character or numeric.

32
 6.7 Solutions 6-101

6.04 Quiz – Correct Answer

Why were the statements and methods that instantiate
and load the hash object inside an IF-THEN/DO group?
data mem_type;
length Code $2 MemberType $40;
if _N_=1 then do;
declare hash T();
T.definekey('Code');
T.definedata('MemberType');
T.definedone();
T.add(key:'10',data:'Orion Club members');
T.add(key:'20',data:'Orion Club Gold members');
T.add(key:'30',data:'Internet/Catalog Customers');
end;
set orion.europe_customers;
rc1=T.find(key:ThisYrType);
if rc1=0 then ThisYrMember=MemberType;
rc2=T.find(key:LastYrType);
if rc2=0 then LastYrMember=MemberType;
run;

The IF-THEN/DO group ensures that the statements

are executed only one time. This saves memory.
41

6.05 Multiple Choice Poll – Correct Answer

What would be the value of LastYrMember if the value of
LastYrType were not found in the hash object (rc2 ne 0)?
a. Missing
b. Orion Club members
c. Orion Club Gold members
d. Internet/Catalog

LastYrMember is initialized to missing at the top of

the DATA step.

67
6-102 Chapter 6 Using DATA Step Hash and Hiter Objects

6.06 Quiz – Correct Answer

Submit the program p306a01 and examine the SAS log.
What are the notes about Code and MemberType?

NOTE: Variable Code is uninitialized.

NOTE: Variable MemberType is uninitialized.

Why do you get those notes?

The DATA step does not provide an initial value for the
variables. The descriptor portion is created by the
LENGTH statement, but it does not provide initial values.
The variables are not used on the left side of the equal
sign (=) in an assignment statement or in a SUM
statement, either of which would provide initial values.

71

6.07 Quiz – Correct Answer

As the last statement in the DO group in the program
p306a01, add the statement:
call missing(Code, MemberType);

Do the notes disappear?

Yes

73
 6.7 Solutions 6-103

6.08 Multiple Choice Poll – Correct Answer

The program p306d02 created the variable rc and then
dropped it. How can you avoid creating the variable so
that you do not have to drop it?
a. Use a WHERE statement or a WHERE= data set
option.
b. Use a KEEP= or DROP= data set option in
orion.product_list.
c. Test the result of the FIND method in the subsetting
IF statement.
d. Use a KEEP or DROP statement.

97

6.09 Quiz – Correct Answer

How do you know the lengths of the character variables
Supplier_Name, Supplier_Address, and Country?

You use PROC CONTENTS, PROC DATASETS, or the

Explorer window to view the descriptor portion of
orion.supplier.

101
6-104 Chapter 6 Using DATA Step Hash and Hiter Objects

6.10 Quiz – Correct Answer

What is the purpose of the BY statement in the DATA
step?

The BY statement creates the variables

FIRST.Customer_ID and LAST.Customer_ID.
LAST.Customer_ID is used to reset the value of
Next_Order_Date to missing when the value of
Customer_ID changes.

153
 6.7 Solutions 6-105

Solutions to Chapter Review

Chapter Review Answers

1. Describe a hash object.
A hash object is used in the DATA step to store
data in memory and to retrieve data from memory.
2. When is a hash object deleted from memory?
When the DATA step completes execution
3. What are the two types of DATA step component
objects?
Hash and hiter
4. What is a key component?
The key component maps key values to data
values.
5. What is the purpose of the DECLARE statement?
To create a hash or hiter object
180

Chapter Review Answers

6. What is the purpose of the FIND method?
The FIND method is used to retrieve data value(s)
based on the value of key(s).
7. What value does the FIND method return when it
executes?
„ Zero when the FIND method finds the KEY value
in the hash object
„ Nonzero when the FIND method does not find
the KEY value in the hash object
8. Are the key(s) and data item(s) variables in the PDV?
Yes, the key(s) and data item(s) must be DATA
step variables in the PDV.

182
6-106 Chapter 6 Using DATA Step Hash and Hiter Objects

Chapter Review Answers

9. Why are the DECLARE, DEFINEKEY, DEFINEDATA,
and DEFINEDONE methods executed in the IF _N_=1
THEN/DO group?
They are executable statements. When they are in
the DO group, they are only executed once per
DATA step.
10. Is the DEFINEDONE method required?
Yes

184
Chapter 7 Creating and Using
Formats

7.1 Using Formats as Lookup Tables ................................................................................. 7-3

Demonstration: Using a Control Data Set to Create a Format .............................................. 7-5

Exercises .............................................................................................................................. 7-21

7.2 Using a Picture Format (Self-Study) ........................................................................... 7-24

Exercises .............................................................................................................................. 7-34

7.3 Chapter Review............................................................................................................. 7-36

7.4 Solutions ....................................................................................................................... 7-37

Solutions to Exercises .......................................................................................................... 7-37

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 7-42

Solutions to Chapter Review ................................................................................................ 7-46

7-2 Chapter 7 Creating and Using Formats
 7.1 Using Formats as Lookup Tables 7-3

7.1 Using Formats as Lookup Tables

Objectives
„ Create permanent formats.
„ Access permanent formats.
„ Create formats from SAS data sets.
„ Maintain formats.
„ Use formats as lookup tables.

Table Lookup Using Formats

The appearance of values is controlled by formats.
„ Use the FORMAT procedure to define tables that store
coded values and the definitions of the codes.
„ Reference these user-defined formats when a table
lookup operation is needed.

4
7-4 Chapter 7 Creating and Using Formats

Overview of a Format (Review)

A format is similar to stacks of buckets that are referred
to by the value of a variable.
Data Value Label
„ SAS puts data values and
label values in the buckets
when the format is used in
a FORMAT statement,
PUT function, or PUT
statement.
„ SAS uses a binary search
on the data value bucket
in order to return the value
in the label bucket.

Business Scenario
The data set orion.country contains the country code
and the country name. Create a format from this data set.

Listing of orion.country
Country_ Country_ Continent_ Country_
Country Population
Name ID ID FormerName
AU Australia 20,000,000 160 96

CA Canada . 260 91
East/West
DE Germany 80,000,000 394 93
Germany
IL Israel 5,000,000 475 95

TR Turkey 70,000,000 905 95

United
US 280,000,000 926 91
States
ZA South Africa 43,000,000 801 94

6
 7.1 Using Formats as Lookup Tables 7-5

Using a Control Data Set to Create a Format

p307d01
/* Step 1 */
/* Make a CNTLIN data set containing */
/* the variables FMTNAME, START, and */
/* LABEL. */
data country;
keep Start Label FmtName;
retain FmtName '$country';
set orion.country(rename=(Country=Start
Country_Name=Label));
run;
proc print data=country noobs;
title 'Country';
run;
/* Step 2 */
/* Use the data set COUNTRY to */
/* make the format $country. */

proc format library=orion.MyFmts cntlin=country fmtlib;

select $country;
run;
/*******************************************************/
/* If there are missing country values, they can be */
/* handled by creating the format $extra. This format */
/* specifically sets missing values to the label */
/* Unknown and uses the label of the $country */
/* format for all other values. Notice that a length */
/* of 30 is provided for the $country format. */
/* The default would be 40. */
/*******************************************************/
proc format library=orion.MyFmts cntlin=country fmtlib;
value $extra ' '='Unknown'
other=[$country30.];
select $country $extra;
title '$country format with missing';
run;

proc catalog cat=orion.MyFmts;

contents;
run;

proc format library=orion.MyFmts fmtlib;

select $country;
run;
7-6 Chapter 7 Creating and Using Formats

Using a Control Data Set to Create a Format

You can create a format from a SAS data set that
contains the code/value information (called a control data
set). Use the CNTLIN= option to read the data and create
the format.
General form of CNTLIN= option:

PROC FORMAT LIBRARY=libref.catalog

CNTLIN=SAS-data-set;
<SELECT format-name format-name...;>
<EXCLUDE format-name format-name...;>
RUN;

The variables FmtName, Start, and Label are

required in order to create a format from a CNTLIN
8 data set.

The CNTLIN= data set has the following features:

• must contain the variables FmtName, Start, and Label
• must contain the variable Type for character formats, unless the value for FmtName begins with a $
• does not require a Type variable for numeric formats
• assumes that the ending value of the format range is equal to the value of Start if no variable named
End is found
• does not require the other variables created by the CNTLOUT= option that specify optional attributes
• can be created by a DATA step, another PROC step, or an interactive application such as the
VIEWTABLE window
• can be used to create new formats, as well as re-create existing formats
• must be grouped by FmtName if multiple formats are specified
 7.1 Using Formats as Lookup Tables 7-7

Setup for the Poll

The DATA step in p307d01 creates the country data set.
data country;
keep Start Label FmtName;
retain FmtName '$country';
set orion.country(rename=(Country=Start
Country_Name=Label));
run;

The DATA step in p307d01a creates an equivalent data

set named country.
data country;
keep Start Label FmtName;
FmtName='$country';
set orion.country;
Start=Country;
Label=Country_Name;
run;

10

7.01 Multiple Choice Poll

Which program should be more efficient?
a. p307d01
b. p307d01a
c. They should be equally efficient.

11
7-8 Chapter 7 Creating and Using Formats

Nesting Formats
In the VALUE statement, you can specify that the format
use a second format as the formatted value.

value=[existing-format]

Enclose the format name in square brackets:

proc format library=orion.MyFmts
cntlin=country fmtlib;
value $extra ' '='Unknown'
other=[$country30.];
select $country $extra;
title '$country format with missing';
run;

p307d01
13

 Avoid nesting formats for more than one level. The resource requirements can increase
dramatically with each additional level.

How Formats Are Stored

Formats are stored as SAS catalog entries.
„ SAS catalogs are special SAS files that store many
different types of information in smaller units called
entries.
„ A single SAS catalog can contain many different
catalog entries.

SAS Catalogs
work.formats orion.formats orion.MyFmts

 Store frequently used formats in permanent catalogs.

14

Catalog entries have four-level names: libref.catalog.entry-name.type.

The type for character formats is formatc. The type for numeric formats is format.
 7.1 Using Formats as Lookup Tables 7-9

Where Formats Are Stored

Without the LIBRARY= option, formats are stored in the
work.formats catalog and exist for the duration of the
SAS session.

PROC FORMAT;

If the LIBRARY= option specifies only a libref, formats

are stored permanently in libref.formats.
PROC FORMAT LIBRARY=libref;

If the LIBRARY= option specifies libref.catalog, formats

are stored permanently in that catalog.

PROC FORMAT LIBRARY=libref.catalog;

15

Documenting Formats
You can use the SAS Explorer Window to view the
formats stored in a catalog.

16
7-10 Chapter 7 Creating and Using Formats

Documenting Formats
The CATALOG procedure manages entries in
SAS catalogs.
Selected capabilities of PROC CATALOG include the
following:
„ creating a listing of the contents of a catalog

„ copying a catalog or selected entries within a catalog

„ renaming or deleting entries within a catalog

„ modifying the description of a catalog entry

17
 7.1 Using Formats as Lookup Tables 7-11

The CATALOG Procedure

proc catalog cat=orion.MyFmts;
contents;
run;

Output
Contents of Catalog ORION.MYFMTS

# Name Type Create Date Modified Date

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
1 DATES FORMAT 29Jan08:16:26:39 29Jan08:16:26:39
2 COUNTRY FORMATC 29Jan08:16:33:30 29Jan08:16:33:30
3 COUNTRY_NAME FORMATC 20Apr09:15:30:14 20Apr09:15:30:14
4 EXTRA FORMATC 20Apr09:15:30:14 20Apr09:15:30:14

p307d01
18

General form of the CATALOG procedure:

PROC CATALOG CATALOG=<libref.>catalog <options>;

CONTENTS <OUT=SAS-data-set> <FILE=fileref>;
COPY OUT=<libref.>catalog <options>;
SELECT entry(s) </ ENTRYTYPE=etype>;
EXCLUDE entry(s) </ ENTRYTYPE=etype>;
DELETE entry(s) </ ENTRYTYPE=etype>;
<RUN;>
QUIT;
7-12 Chapter 7 Creating and Using Formats

Documenting Formats
You can use the FMTLIB option in the PROC FORMAT
statement to document the format.
proc format library=orion.MyFmts fmtlib;
select $country;
run;

„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚ FORMAT NAME: $COUNTRY LENGTH: 13 NUMBER OF VALUES: 7 ‚
‚ MIN LENGTH: 1 MAX LENGTH: 40 DEFAULT LENGTH 13 FUZZ: 0 ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚START ‚END ‚LABEL (VER. V7|V8 05MAY2009:12:34:42)‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚AU ‚AU ‚Australia ‚
‚CA ‚CA ‚Canada ‚
‚DE ‚DE ‚Germany ‚
‚IL ‚IL ‚Israel ‚
‚TR ‚TR ‚Turkey ‚
‚US ‚US ‚United States ‚
‚ZA ‚ZA ‚South Africa ‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒŒ

p307d01
19

General form of the FMTLIB option:

PROC FORMAT LIBRARY=libref.catalog FMTLIB;

<SELECT format-name format-name...;>
<EXCLUDE format-name format-name...;>
RUN;

 You can use either the SELECT or EXCLUDE statement to process specific formats rather than
an entire catalog.

Using Formats
You can reference formats in any of the following:
„ FORMAT statements

„ PUT statements

„ PUT functions in assignment, WHERE,

or IF statements
„ FORMAT= options

20
 7.1 Using Formats as Lookup Tables 7-13

Using Formats
When a format is referenced, SAS does the following:
„ loads the format from the catalog entry into memory

„ performs a binary search on values in the table

to execute a lookup
„ returns a single result for each lookup

21

7.02 Quiz
Submit the program p307a01.
What error messages do you see in the SAS log?

data customers;
set orion.customer;
Country_Name=put(Country,$country.);
run;

proc freq data=orion.employee_addresses;

tables Country;
format Country $extra.;
run;

23
7-14 Chapter 7 Creating and Using Formats

7.03 Quiz
1. Add the following OPTIONS statement to p307a01
and resubmit the program. What is the result?
options nofmterr;

2. Replace the current OPTIONS statement with the

following statement and resubmit the program.
What is the result?

options fmterr fmtsearch=(orion orion.MyFmts);

25

Using the NOFMTERR System Option

By default, the FMTERR system option is in effect. If you
use a format that SAS cannot load, SAS issues an error
message and stops processing the step.
To prevent the default action, change the system option
FMTERR to NOFMTERR.

OPTIONS FMTERR | NOFMTERR;

28

FMTERR specifies that when SAS cannot find a specified variable format, it
generates an error message and does not allow default substitution to occur.

NOFMTERR replaces missing formats with the w. or $w. default format, issues a note,
and continues processing.
 7.1 Using Formats as Lookup Tables 7-15

Using the FMTSEARCH= System Option

To use permanent formats or to search multiple catalogs,
use the FMTSEARCH= system option to identify the
catalog(s) to be searched for the format(s).
General form of the FMTSEARCH= system option:

OPTIONS FMTSEARCH=(item-1 item-2…item-n);

29

Using the FMTSEARCH= System Option

options fmtsearch=(orion orion.MyFmts);

SAS Supplied Formats

work.formats

library.formats

orion.formats

orion.MyFmts

30

Because orion is a libref without a catalog name, formats is assumed to be the catalog name.
SAS supplied formats are always searched first. The work.formats catalog is always searched second,
unless it appears in the FMTSEARCH list. If the library libref is assigned, the library.formats catalog is
searched after work.formats and before anything else in the FMTSEARCH list, unless it appears in the
list. To assign the library libref, use the code shown below:

libname library 'SAS-data-library-containing-format-catalog';

7-16 Chapter 7 Creating and Using Formats

Maintaining Formats
To maintain formats, perform one of the following tasks:
„ Edit the PROC FORMAT code that created the original
format.
„ Create a SAS data set from the format, edit the data
set, and use the CNTLIN= option to re-create the
format.

31

Maintaining Permanent Formats

Step 1
proc format library=libref.catalog
Permanent cntlout=SAS-data-set;
Formats select format-name;
Catalog run;

Step 2
SAS Edit
Data Set Values
Step 3
proc format library=libref.catalog
cntlin=SAS-data-set;
run;

32

 When the data set created by the CNTLOUT= option will be used as a CNTLIN= data set in a
subsequent FORMAT procedure step, the minimum variables that must be included are START,
END, FMTNAME, and LABEL.
 7.1 Using Formats as Lookup Tables 7-17

Maintaining Permanent Formats

p307d02
/* Step 1 */

proc format library=orion.MyFmts cntlout=countryfmt;

select $country;
run;

proc print data=countryfmt;

run;

/* Step 2 */

proc sql;
insert into countryfmt(FmtName, Start, End, Label)
values('$country', 'BR', 'BR', 'Brazil')
values('$country', 'CH', 'CH', 'Switzerland')
values('$country', 'MX', 'MX', 'Mexico');
quit;

/* Step 3 */

proc format library=orion.MyFmts cntlin=countryfmt fmtlib;

select $country;
run;

/* to add Missing in the $extra Format */

proc format library=orion.MyFmts cntlin=countryfmt fmtlib;

value $extra ' '='Unknown'
other=[$country30.];
select $country $extra;
title '$country format with missing';
run;
You can use either the SELECT or EXCLUDE statement to process specific formats rather than an entire
catalog.
The variables in the output control data set completely describe all aspects of each format or informat,
including optional settings. The output control data set contains one observation per range per format or
informat in the specified catalog.
7-18 Chapter 7 Creating and Using Formats

The CNTLOUT data set contains the following variables:

DEFAULT a numeric variable that indicates the default length for format or informat

END a character variable that gives the range’s ending value

EEXCL a character variable that indicates whether the range’s ending value is excluded

FILL for picture formats, a numeric variable whose value is the value of the FILL= option

FMTNAME a character variable whose value is the format or informat name

FUZZ a numeric variable whose value is the value of the FUZZ= option

HLO a character variable that contains range information about the format or informat in
the form of different letters that can appear in any combination

LABEL a character variable whose value is the informatted or formatted value or the name of
an existing informat or format

LENGTH a numeric variable whose value is the value of the LENGTH= option

MAX a numeric variable whose value is the value of the MAX= option

MIN a numeric variable whose value is the value of the MIN= option

MULT a numeric variable whose value is the value of the MULT= option

NOEDIT for picture formats, a numeric variable whose value indicates whether the NOEDIT
option is in effect

PREFIX for picture formats, a character variable whose value is the value of the PREFIX=
option

SEXCL a character variable that indicates whether the range’s starting value is excluded

START a character variable that gives the range’s starting value

TYPE a character variable that indicates the type of format

 7.1 Using Formats as Lookup Tables 7-19

Maintaining Permanent Formats

General form of PROC FORMAT with the CNTLOUT=
option:

PROC FORMAT LIBRARY=libref.catalog

CNTLOUT=SAS-data-set;
<SELECT format-name format-name...;>
<EXCLUDE format-name format-name...;>
RUN;

34

Advantages and Disadvantages of Formats

Advantages Disadvantages
familiarity memory requirements to load the
entire format for the binary search
no need to create additional data use of only one variable for the table
lookup
ability to be used with procedures requirement of more disk space to
store a format than to store the
equivalent SAS data set
range search for both character and
numeric
binary search through lookup table
centralized maintenance
use of multiple PUT functions to
create multiple variables

35
ability to be stored permanently
7-20 Chapter 7 Creating and Using Formats

Comparing Arrays, Hash Objects, and Formats

Array Hash Object Format
The subscript value(s) The keys can be character, Ranges can be character or
must be numeric. numeric, or both. numeric.
One data value can be Multiple data items can be Only one data value can be
associated with the associated with the key associated with a label.
subscript value(s). value.
An array uses less memory A hash object uses more A format requires more
than a hash object. memory than an array. memory than a hash object.
The size of the array is The size of the hash object The memory requirement for
determined at compilation is determined at execution the format is determined when
time. time. the format is used.
Subscript values must be The keys do not have to be Formats support mapping a
consecutive integers. consecutive or sorted. range of values or a list of
discrete values to one value.
An array selects values by A hash object uses a hash A format uses a binary search
direct access based on the function for the lookup for the lookup process.
subscript value. process.
Arrays can only be used in Hash objects can only be Formats can be used in DATA
the DATA step. used in the DATA step. and PROC steps.
36

To estimate the amount of memory used by a format, refer to Usage Note 23084 at
support.sas.com/kb/23/084.html.
 7.1 Using Formats as Lookup Tables 7-21

Exercises

Level 1

1. Creating Formats with Values from a SAS Data Set

The data set orion.continent contains the Continent_ID and the Continent_Name variables.
Continent_
Obs ID Continent_Name

1 91 North America
2 93 Europe
3 94 Africa
4 95 Asia
5 96 Australia/Pacific

a. Create a CNTLIN data set named continent that reads the data from orion.continent and contains
the variables FmtName, Start, and Label. The name of the format should be CONTINENT.
b. Use the CNTLIN= option to create a format from the continent data set and store the format in
the orion.MyFmts catalog.
c. Open the program p307e01c and submit it. The program should execute successfully with no
errors in the SAS log.
p307e01c
/*******************/
/* Part C */
/* Use continent. */
/*******************/

data countries;
set orion.country;
Continent_Name=put(Continent_ID, continent.);
run;

proc print data=countries(obs=10);

title 'Continent Names';
run;
7-22 Chapter 7 Creating and Using Formats

d. Open the program p307e01d.

p307e01d
/*************************************************/
/* When START and END are created using the */
/* CNTLOUT= option, they are created as character*/
/*************************************************/

proc sql;
insert into continentfmt(fmtname, Start, End, Label)
values('continent', '90', '90', 'Antarctica')
values('continent', '92', '92', 'South America');
quit;
1) Before the PROC SQL step, add a PROC FORMAT step with the CNTLOUT= option to
create a control output data set named continentfmt from the CONTINENT format.
2) Submit the program to add new observations to the continentfmt data set.
3) Add another PROC FORMAT step with the CNTLIN= option to read the continentfmt data
set and re-create the CONTINENT format. Use the FMTLIB option in this PROC FORMAT
step to ensure that the new values were added to the format CONTINENT.

Level 2

2. Creating Formats with Inclusive Ranges from a SAS Data Set

The data set orion.ages contains three variables: First_Age, Last_Age, and Description.
Partial Listing of orion.ages
First_
Obs Age Last_Age Description

1 15 30 15-30 years
2 30 45 31-45 years
3 45 60 46-60 years
4 60 75 61-75 years

a. Create a format from the orion.ages data set and store it permanently in the orion.MyFmts
catalog. Use the appropriate option to view the values in the format.
b. Write a DATA step to create a data set named sales that reads the Employee_ID and Birth_Date
variables from the orion.sales data set. Create a new variable named Age that is the employee’s
age as of the current date and another new variable named Age_Cat that is the value of the
variable Age using the AGE format.
 7.1 Using Formats as Lookup Tables 7-23

c. Print the first five observations of the sales data set to confirm that the new variables were created
correctly.
PROC PRINT Output (As of May 5, 2009)
Sales Data Set

Birth_
Obs Employee_ID Date Age Age_Cat

1 120102 11AUG1969 39 31-45 years

2 120103 22JAN1949 60 46-60 years
3 120121 02AUG1944 64 61-75 years
4 120122 27JUL1954 54 46-60 years
5 120123 28SEP1964 44 31-45 years

Level 3

3. Creating Formats with Exclusive Ranges from a SAS Data Set

The data set orion.ages_mod contains three variables: First_Age, Last_Age, and Description.
Partial Listing of orion.ages_mod
First_
Obs Age Last_Age Description

1 15 30 15-29 years
2 30 45 30-44 years
3 45 60 45-59 years
4 60 75 60-75 years

a. Create a format named AGES_MOD from the orion.ages_mod data set and store it permanently
in the orion.MyFmts catalog. Use the appropriate option to view the values in the format.

 The value of the Last_Age variable is not to be included in the Description variable. Use
SAS Help or SAS OnlineDoc to investigate the EEXCL variable that is required to get
the correct results for this exercise.
b. Write a DATA step to create a data set named sales that reads the Employee_ID and Birth_Date
variables from the orion.sales data set. Create a new variable named Age that is the employee’s
age as of the current date and another new variable named Age_Cat that is the value of the
variable Age using the AGES_MOD format.
c. Print the first five observations of the sales data set to confirm that the new variables were created
correctly.
PROC PRINT Output (as of May 5, 2009)
Sales Data Set

Birth_
Obs Employee_ID Date Age Age_Cat

1 120102 11AUG1969 39 30-44 years

2 120103 22JAN1949 60 45-59 years
3 120121 02AUG1944 64 60-75 years
4 120122 27JUL1954 54 45-59 years
5 120123 28SEP1964 44 30-44 years
7-24 Chapter 7 Creating and Using Formats

7.2 Using a Picture Format (Self-Study)

Objectives
„ Use a picture format to format numeric data.

40

Using Picture Formats

Formatting numeric data often involves inserting special
characters into the numeric values.
The following illustrates the original numeric value and
the required formatted value.
Original Numeric Formatted Value
Data Value
5552134567 (555)213-4567
25 25%
-25.12 25.12DR

213 **********213

41
 7.2 Using a Picture Format (Self-Study) 7-25

Picture Formats
Some uses for picture formats include the following:
„ displaying numbers with leading zeros (0005)

„ filling numbers with special characters (***5)

„ inserting message characters into numbers, such

as parentheses (for phone numbers), percent signs,
and minus signs
„ customizing a date, time, or datetime display with
directives
„ formatting currency when there is no SAS format
available

 Picture formats consist of placeholders for digits and

special characters to be inserted into the digits.
42

Business Scenario
The data set orion.phone contains the phone number of
the employees from the United States and Australia. The
phone number is stored in a numeric variable named
Phone. Create a data set that contains the phone number
in the correct formatted form.
Partial Listing of orion.phone
Employee_
Phone_Type Country Phone
ID
120101 Home AU 61255551849
120101 Work AU 61255510001
120102 Home AU 61355559700
. . . .
. . . .
. . . .
121147 Home US 13055510423
121148 Work US 13055554118
121148 Home US 13055510424
43
7-26 Chapter 7 Creating and Using Formats

Using a PICTURE Format

proc format;
picture us_phone
low-high='9 (999) 999-9999';
picture au_phone
low-high='99 (9) 9999-9999';
run;

data phone_list;
set orion.phone;
if Country='AU' then
Phone_Number=put(Phone,au_phone.);
else if Country='US' then
Phone_Number=put(Phone,us_phone.);
run;

p307d03
44

7.04 Quiz
Open and submit the program p307d03.
1. How are the Australian phone numbers displayed?

2. How are the United States phone numbers displayed?

46
 7.2 Using a Picture Format (Self-Study) 7-27

Using a PICTURE Format

Partial Output
Phone Numbers Formatted

Employee_ Phone_
Obs ID Type Country Phone Phone_Number

1 120101 Home AU 61255551849 61 (2) 5555-1849

2 120101 Work AU 61255510001 61 (2) 5551-0001
3 120102 Home AU 61355559700 61 (3) 5555-9700
4 120102 Work AU 61355510002 61 (3) 5551-0002
5 120103 Home AU 61255553998 61 (2) 5555-3998

<observations removed>

913 121146 Work US 12155550546 1 (215) 555-0546

914 121146 Home US 12155510422 1 (215) 551-0422
915 121147 Work US 13055555653 1 (305) 555-5653
916 121147 Home US 13055510423 1 (305) 551-0423
917 121148 Work US 13055554118 1 (305) 555-4118
918 121148 Home US 13055510424 1 (305) 551-0424

49

The PICTURE Statement

The PICTURE statement in PROC FORMAT defines
a pattern for data values.

PROC FORMAT;
PICTURE name
value-or-range-1 <..., value-or-range-n>='picture';
RUN;

50
7-28 Chapter 7 Creating and Using Formats

Value Ranges in the PICTURE Statement

PICTURE name range='picture' <options>;

The picture consists of the following:

„ Digit selectors that are numeric characters (0 through
9) that define positions for numeric values
If you use a digit selector of 0, the format suppresses
leading zeros in variable values. If you use a digit
selector of 1 through 9, then the format pattern
displays numbers that are padded on the left with
zeros.
„ Message characters that are nonnumeric characters
that print as specified in the picture
Using the PREFIX= option, specify nonnumeric
characters to be printed before the numeric digits.
51

Using a PICTURE Format

proc format;
picture us_phone
low-high='9 (999) 999-9999';
picture au_phone
low-high='99 (9) 9999-9999';
run;

Variable: Country Variable: Phone Variable:

Phone_Number
AU 61255551849 61 (2) 5555-1849
US 12155555906 1 (215) 555-5906

p307d03
52
 7.2 Using a Picture Format (Self-Study) 7-29

Using a PICTURE Format

proc format;
picture us_phone
low-high='9 (999) 999-9999';
picture us_phone_withzeros
low-high='0 (000) 000-9999';
run;

Phone Displayed with Displayed with

us_phone. us_phone_withzeros.
12155555906 1 (215) 555-5906 1 (215) 555-5906

2155555906 0 (215) 555-5906 215) 555-5906

5555906 0 (000) 555-5906 555-5906

5906 0 (000) 000-5906 5906

p307d04
53

To insert the open parenthesis in the phone number 2155555906, use the following PROC FORMAT step:
proc format;
picture us_phone_withzeros
low-<10000000='0 (000) 000-9999'
10000000-high='0 (000) 000-9999' (prefix='(');
run;

Using Digit Selectors

proc format;
picture padz low - high='999,999.99';
picture nopad low - high='000,000.00';
picture mixed low – high='000,009.99';
run;

Displayed Displayed Displayed All data

Variable
with with with values
Value
padz. nopad. mixed. treated as
2381.6 002,381.60 2,381.60 2,381.60 positive
values
-12.233 000,012.23 12.23 12.23
Decimal not
.38 000,000.38 38 0.38 used with
12345 012,345.00 12,345.00 12,345.00 zero digit
selectors
123456 123,456.00 123,456.00 123,456.00

45.987 000,045.98 45.98 45.98 p307d05

54
7-30 Chapter 7 Creating and Using Formats

7.05 Quiz
Submit the program p307a02. How many digits are
displayed to the right of the decimal point?
proc format;
picture rtfmt 0 - high='999,999.9999';
picture wzrfmt 0 - high='000,009.0000';
run;

56

7.06 Quiz
Submit the program p307a03. How many digits are
printed to the left of the decimal point?
proc format;
picture small low - high='0,009.99';
picture large low - high='000,009.99';
run;

58
 7.2 Using a Picture Format (Self-Study) 7-31

Picture Format Examples

Data Picture Display
Values Format As…
-12.233 bank low-<0='009.00DR' 12.23DR
0-high='009.00CR'
-12.233 sign low-<0='009.00' (prefix="-") -12.23
0-high='009.00'
45.987 numrd(round) other= '000,009.99' 45.99

24.319 pctrnd(round) low-high=' 009.99%' 24.32%

24.319 pct low-high=' 009.99%' 24.31%
getdec
.38 0-<1='99' (prefix='.' mult=100) .38
1.38 1-high='00.99'; 1.38
p307d06
60

Selected Options in the PICTURE Statement

To Do This… Use This…
Specify a number to multiply the variable's value MULTIPLIER=
by before it is formatted
Specify multiple pictures for a given value or MULTILABEL
range and for overlapping ranges
Store values or ranges in the order in which you NOTSORTED
define them
Round the value to the nearest display value ROUND
before formatting
Specify a character that completes the formatted FILL=
value
Specify a character prefix for the formatted value PREFIX=

61
7-32 Chapter 7 Creating and Using Formats

Inserting Characters
proc format;
picture paren low - high='(999)999-9999';
picture nospace low-high='999)999-9999'
(prefix= '(' );
picture space low-high=' 999)999-9999'
(prefix= '(' );
run;

Value Displayed with Displayed with Displayed with

paren. nospace. space.

8005550202 800)555-0202 800)555-0202 (800)555-0202

You must ensure that the length of the format is large

enough for the prefix characters and all the possible digits
in your variable value. Otherwise, the formatted value
might not appear properly.
p307d07
62

Date, Time, and Datetime Directives

Using special directives, it is possible to specify a picture
format for extensive combinations of dates and times.
„ Date, time, and datetime directives start with a
percent sign (%).
„ Directives are case sensitive.

„ Because of the % in the directive, picture format labels

that include these directives should be enclosed in
single quotation marks.
„ If you intend to use the date, time, or datetime
directives, you must also specify the DATATYPE=
option in the PICTURE statement.
PICTURE name
range='picture' (DATATYPE=DATE | TIME | DATETIME);
63
 7.2 Using a Picture Format (Self-Study) 7-33

Date Directives
Consider this date value: -3334 (November 15, 1950).
Use This Variable
To Display
Directive Display
Abbreviated weekday name %a Wed
Full weekday name %A Wednesday
Abbreviated month name %b NOV
Full month name %B November
Month value as decimal %m 11

Day of the month (decimal) %d 15

Year without century %y 50

Year with century %Y 1950

64

Date Directives
proc format;
picture longdate (default=30)
'01jan1950'd-'31dec2004'd='%A, %B %d'
(datatype=date);
picture noleadz
'01jan1950'd-'31dec2004'd='%y~%m~%d'
(datatype=date);
picture leadzero
'01jan1950'd-'31dec2004'd='%0y~%0m~%0d'
(datatype=date);
run;

SAS Date Displayed Displayed Displayed

Value with longdate. with noleadz. with leadzero.

-3334 Wednesday, November 15 50~11~15 50~11~15

15101 Sunday, May 6 1~5~6 01~05~06

p307d08
65
7-34 Chapter 7 Creating and Using Formats

Exercises

Level 1

4. Using a PICTURE Format for Numeric Data

a. Write a PROC FORMAT step with the PICTURE statement to format the value of the variable
Order_ID so that it appears as follows:

Variable Value Formatted Value

1230058123 12-30-05-8123

b. Print the first five observations of the orion.order_fact data set to validate the formatted values
of the variable.
PROC PRINT Output
Formatted Values of Order_ID

Obs Order_ID

1 12-30-05-8123
2 12-30-08-0101
3 12-30-10-6883
4 12-30-14-7441
5 12-30-31-5085

Level 2

5. Using a PICTURE Format for Currency Data

a. Write a PROC FORMAT step with the PICTURE statement to format the value of the variable
Total_Revenue_Price so that it appears as follows:

Variable Value Formatted Value

27.80 kr. 27,80 eks.moms

 7.2 Using a Picture Format (Self-Study) 7-35

b. Print the first five observations of the orion.denmark_customers data set to validate the
formatted values of the variable.
PROC PRINT Output
Using a PICTURE Format

Obs Total_Retail_Price

1 kr. 27,80 eks.moms

2 kr. 196,20 eks.moms
3 kr. 97,60 eks.moms
4 kr. 65,60 eks.moms
5 kr. 138,20 eks.moms

Additional information about the Danish Kroner formatting:

Abbreviation Meaning
eks eksklusiv (exclusive)

moms MerOMsætningsafgift (value added tax)

Level 3

6. Using a PICTURE Format for Date Variables

a. Write a PROC FORMAT step with the PICTURE statement to format the value of the variable
Order_Date so that it appears as follows:

Variable Value Formatted Value

15716 Saturday, 1.11.2003

b. Print the first five observations of the orion.order_fact data set to validate the formatted values
of the variable.
PROC PRINT Output
Obs Order_Date

1 Saturday, 1.11.2003
2 Wednesday, 1.15.2003
3 Monday, 1.20.2003
4 Tuesday, 1.28.2003
5 Thursday, 2.27.2003
7-36 Chapter 7 Creating and Using Formats

7.3 Chapter Review

Chapter Review
1. What PROC FORMAT statement option is used
to create a permanent format?

2. What PROC FORMAT statement option is used

to create a format from a SAS data set?

3. What variables are required to create a format from

a SAS data set?

68 continued...

Chapter Review
4. What PROC FORMAT option is used to view
the contents of a format?

5. What PROC FORMAT option is used to create

a SAS data set from a format?

6. What is the SAS system option that establishes

the search path when a format is used?

70
 7.4 Solutions 7-37

7.4 Solutions

Solutions to Exercises
1. Creating Formats with Values from a SAS Data Set
a. Create a CNTLIN data set named continent that reads the data from orion.continent and contains
the variables FmtName, Start, and Label. The name of the format should be CONTINENT.
p307s01
/*********************/
/* Part A */
/* Make continent */
/*********************/

data continent;
keep Start Label FmtName;
retain FmtName 'continent';
set orion.continent(rename=(Continent_ID=Start
Continent_Name=Label));
run;

proc print data=continent(obs=10) noobs;

title 'Continent';
run;
b. Use the CNTLIN= option to create a format from the continent data set and store the format in
the orion.MyFmts catalog.
p307s01
/***********************************/
/* Part B */
/* Use continent to create format. */
/***********************************/

proc format library=orion.MyFmts cntlin=continent fmtlib;

select continent;
title 'Continent format';
run;
c. Open the program p307e01c and submit it. The program should execute successfully with no
errors in the SAS log.
7-38 Chapter 7 Creating and Using Formats

d. Open the program p307e01d.

1) Before the PROC SQL step, add a PROC FORMAT step with the CNTLOUT= option to
create a control output data set named continentfmt from the CONTINENT format.
p307s01
/********************/
/* Part D */
/* Update Continent.*/
/********************/
proc format library=orion.MyFmts cntlout=continentfmt;
select continent;
run;
2) Submit the program to add new observations to the continentfmt data set.
3) Add another PROC FORMAT step with the CNTLIN= option to read the continentfmt data
set and re-create the CONTINENT format. Use the FMTLIB option in this PROC FORMAT
step to ensure that the new values are added to the CONTINENT format.
p307s01
proc format library=orion.MyFmts cntlin=continentfmt
fmtlib;
select continent;
run;
2. Creating Formats with Inclusive Ranges from a SAS Data Set
a. Create a format from the orion.ages data set and store it permanently in the orion.MyFmts
catalog. Use the appropriate option to view the values in the format.
p307s02
data ages;
set orion.ages (rename=(First_Age=Start Last_Age=End
Description=Label));
retain FmtName 'ages';
run;
proc format library=orion.MyFmts fmtlib cntlin=ages;
select ages;
run;
b. Write a DATA step to create a data set named sales that reads the Employee_ID and Birth_Date
variables from the orion.sales data set. Create a new variable named Age that is the employee’s
age as of the current date and another new variable named Age_Cat that is the value of the
variable Age using the AGE format.
p307s02
data sales;
set orion.sales(keep=Employee_ID Birth_Date);
Age=int(yrdif(Birth_Date, today(), 'ACT/ACT'));
Age_Cat=put(Age, ages.);
run;
 7.4 Solutions 7-39

c. Print the first five observations of the sales data set to confirm that the new variables were created
correctly.
p307s02
proc print data=sales(obs=5);
format Birth_Date date9.;
title 'Sales Data Set';
run;
3. Creating Formats with Exclusive Ranges from a SAS Data Set
a. Create a format named ages_mod from the orion.ages_mod data set and store it permanently in
the orion.MyFmts catalog. Use the appropriate option to view the values in the format.
p307s03
data ages_mod;
set orion.ages_mod(rename=(First_Age=Start Last_Age=End
Description=Label));
retain fmtname 'ages_mod';
EEXCL='Y';
run;

proc format library=orion.MyFmts fmtlib cntlin=ages_mod;

select ages_mod;
run;
b. Write a DATA step to create a data set named sales that reads the Employee_ID and Birth_Date
variables from the orion.sales data set. Create a new variable named Age that is the employee’s
age as of the current date, and another new variable named Age_Cat that is the value of the
variable Age using the AGES_MOD format.
p307s03
options fmtsearch=(orion.MyFmts);

data sales;
set orion.sales(keep=Employee_ID Birth_Date);
Age=int(yrdif(Birth_Date, today(), 'ACT/ACT'));
Age_Cat=put(Age, ages_mod.);
run;
c. Print the first five observations of the sales data set to confirm that the new variables were created
correctly.
p307s03
proc print data=sales(obs=5);
format birth_date date9.;
title 'Sales Data Set';
run;
7-40 Chapter 7 Creating and Using Formats

4. Using a PICTURE Format for Numeric Data

a. Write a PROC FORMAT step with the PICTURE statement to format the value of the variable
Order_ID so that it appears as follows:

Variable Value Formatted Value

1230058123 12-30-05-8123

b. Print the first five observations of the orion.order_fact data set to validate the formatted values
of the variable.
p307s04
proc format;
picture product low – high='99-99-99-9999';
run;

proc print data=orion.order_fact(obs=5);

format Order_ID product.;
var Order_ID;
title "Formatted Values of Order_ID";
run;
5. Using a PICTURE Format for Currency Data
a. Write a PROC FORMAT step with the PICTURE statement to format the value of the variable
Total_Revenue_Price so that it appears as follows:

Variable Value Formatted Value

27.80 kr. 27,80 eks.moms

b. Print the first five observations of the orion.denmark_customers data set to validate the
formatted values of the variable.
p307s05
proc format;
picture kroner 0 - high='000.009,99 eks.moms' (mult=100
prefix='kr. ');
run;

proc print data=orion.denmark_customers(obs=5);

format Total_Retail_Price kroner.;
title 'Using a PICTURE Format';
var Total_Retail_Price;
run;
 7.4 Solutions 7-41

6. Using a PICTURE Format for Date Variables

a. Write a PROC FORMAT step with the PICTURE statement to format the value of the variable
Order_Date so that it appears as follows:

Variable Value Formatted Value

15716 Saturday, 1.11.2003

b. Print the first five observations of the orion.order_fact data set to validate the formatted values
of the variable.
p307s06
proc format;
picture day_of_week(default=21) low – high='%A, %m.%d.%Y'
(datatype=date);
run;

proc print data=orion.order_fact(obs=5);

title 'Day of Week Format';
format Order_Date day_of_week.;
var Order_Date;
run;
7-42 Chapter 7 Creating and Using Formats

Solutions to Student Activities (Polls/Quizzes)

7.01 Multiple Choice Poll – Correct Answer

Which program should be more efficient?
a. p307d01
b. p307d01a
c. They should be equally efficient.

12

7.02 Quiz – Correct Answer

Submit the program p307a01.
What error messages do you see in the SAS log?
477 data customers;
478 set orion.customer;
479 Country_Name=put(Country,$country.);
---------
48
ERROR 48-59: The format $COUNTRY was not found or could not be loaded.

480 run;

NOTE: The SAS System stopped processing this step because of errors.
WARNING: The data set WORK.CUSTOMERS may be incomplete. When this step was stopped there were
0 observations and 13 variables.
WARNING: Data set WORK.CUSTOMERS was not replaced because this step was stopped.
NOTE: DATA statement used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

481
482 proc freq data=orion.employee_addresses;
483 tables Country;
484 format Country $extra.;
ERROR: The format $EXTRA was not found or could not be loaded.
485 run;

NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE FREQ used (Total process time):
real time 0.10 seconds
cpu time 0.00 seconds
24
 7.4 Solutions 7-43

7.03 Quiz – Correct Answer

1. Add the following OPTIONS statement to p307a01
and resubmit the program. What is the result?
options nofmterr;
All of the procedure steps were executed with no
warnings or errors in the SAS log. The user-
defined formats were not applied.
2. Replace the current OPTIONS statement with the
following statement and resubmit the program.
What is the result?
options fmterr fmtsearch=(orion orion.MyFmts);
All of the procedure steps were executed with no
warnings or errors in the SAS log. The user-
defined formats were applied.
27

Modified p307a01s for Quiz 7.03, Step 1

options nofmterr;

data customers;
set orion.customer;
Country_Name=put(Country,$country.);
run;

proc freq data=orion.employee_addresses;

tables Country;
format Country $extra.;
run;
Modified p307a01s for Quiz 7.03. Step 2
options fmtsearch=(orion orion.MyFmts);

data customers;
set orion.customer;
Country_Name=put(Country,$country.);
run;

proc freq data=orion.employee_addresses;

tables Country;
format Country $extra.;
run;
7-44 Chapter 7 Creating and Using Formats

7.04 Quiz – Correct Answer

Open and submit the program p307d03.
1. How are the Australian phone numbers displayed?
Two digits, a blank, one digit in parentheses,
a blank, four digits, a dash, four digits.
61 (2) 5555-1849

2. How are the United States phone numbers displayed?

The digit 1, a blank, three digits in parentheses,
a blank, three digits, a dash, four digits:
1 (215) 555-0546

48

7.05 Quiz – Correct Answer

Submit the program p307a02. How many digits are
displayed to the right of the decimal point?
proc format;
picture rtfmt 0 - high='999,999.9999';
picture wzrfmt 0 - high='000,009.0000';
run;
Variable Displayed with Displayed with
Value rtfmt. wzrfmt. No matter which
.5 000,000.5000 0.5000 digit selector
you specify, all
1.5 000,001.5000 1.5000
digits to the
12.55 000,012.5500 12.5500 right of the
123.555 000,123.5550 123.5550 decimal place
1234.5555 001,234.5555 1,234.5555 are displayed.

57
 7.4 Solutions 7-45

7.06 Quiz – Correct Answer

Submit the program p307a03. How many digits are
printed to the left of the decimal point?
proc format;
picture small low - high='0,009.99';
picture large low - high='000,009.99';
run;

Variable Displayed with Displayed with If the variable value

Value small. large. is bigger than the
1234.56 1,234.56 1,234.56 PICTURE format,
12345.67 2,345.67 12,345.67 only the rightmost
123456.78 3,456.78 123,456.78 digits are
displayed.
1234567.89 4,567.89 234,567.89

59
7-46 Chapter 7 Creating and Using Formats

Solutions to Chapter Review

Chapter Review Answers

1. What PROC FORMAT statement option is used
to create a permanent format?
LIBRARY=
2. What PROC FORMAT statement option is used
to create a format from a SAS data set?
CNTLIN=
3. What variables are required to create a format from
a SAS data set?
FMTNAME, START, and LABEL, at a minimum.
The END variable is required if the SAS data set
was created using the CNTLOUT= option.

69 continued...

Chapter Review Answers

4. What PROC FORMAT option is used to view
the contents of a format?
FMTLIB
5. What PROC FORMAT option is used to create
a SAS data set from a format?
CNTLOUT=
6. What is the SAS system option that establishes
the search path when a format is used?
FMTSEARCH=(library-1, library-2, . . . , library-n)

71
Chapter 8 Combining Data
Horizontally

8.1 DATA Step Merges and SQL Procedure Joins ............................................................. 8-3
Demonstration: Using the DATA Step to Perform a Match-Merge ......................................... 8-7

Demonstration: Using a PROC SQL Join to Perform a Match-Merge ................................. 8-10

Exercises (Optional) ............................................................................................................. 8-17

8.2 Using an Index to Combine Data................................................................................. 8-22

Demonstration: Using Multiple SET … KEY= Statements (Self-Study) ............................... 8-40

Exercises .............................................................................................................................. 8-42

8.3 Combining Summary and Detail Data ......................................................................... 8-49

Exercises .............................................................................................................................. 8-64

8.4 Combining Data Conditionally (Self-Study) ............................................................... 8-71

Demonstration: Using a Hash Object................................................................................... 8-90

Exercises .............................................................................................................................. 8-92

8.5 Chapter Review............................................................................................................. 8-96

8.6 Solutions ....................................................................................................................... 8-98

Solutions to Exercises .......................................................................................................... 8-98

Solutions to Student Activities (Polls/Quizzes) ................................................................... 8-116

Solutions to Chapter Review .............................................................................................. 8-123

8-2 Chapter 8 Combining Data Horizontally
 8.1 DATA Step Merges and SQL Procedure Joins 8-3

8.1 DATA Step Merges and SQL Procedure Joins

Objectives
„ Use the DATA step with a MERGE statement
to combine more than two SAS data sets.
„ Use the SQL procedure to join SAS data sets
without a common variable.
„ Describe the differences between the DATA step
MERGE statement and PROC SQL.

Methods for the Match-Merge

You can perform a match-merge of two or more SAS data
sets with the following techniques:
„ DATA step with the MERGE statement and a BY
statement
„ PROC SQL join

4
8-4 Chapter 8 Combining Data Horizontally

The DATA Step Merge (Review)

When data sets are combined using the DATA step
MERGE statement, both the matches and the
nonmatches on the BY variable are included in the results.
data three;
merge one two;
by X;
run;
three
two X Y Z
one X Z
1 a f
X Y 1 f
1 d r
1 a 1 r 1 d s
1 d 1 s 3 t
4 c 3 t 4 c w
5 g 4 w 5 g

8.01 Multiple Choice Poll

By default, how does the DATA step perform a merge?
a. sequentially
b. creates a Cartesian product

7
 8.1 DATA Step Merges and SQL Procedure Joins 8-5

Business Scenario
The SAS data set orion.staff contains the employee’s ID
and the employee’s manager’s ID.
Partial Listing of orion.staff
Emp_
Employee_ Emp_Hire_ Manager
Start_Date . . . Term_
ID Date _ID
Date
120101 01JUL2003 ... 01JUL2003 . 120261
120102 01JUN1989 ... 01JUN1989 . 120101
120103 01JAN1974 ... 01JAN1974 . 120101
120104 01JAN1981 ... 01JAN1981 . 120101
120105 01MAY1999 ... 01MAY1999 . 120101
120106 01JAN1974 ... 01JAN1974 . 120104
120107 01FEB1974 ... 01FEB1974 . 120104
120108 01AUG2006 ... 01AUG2006 . 120104
. . . . . .
. . . . . .
9 . . . . . .

Business Scenario
The SAS data set orion.employee_addresses
contains the employee’s ID and name.
Partial Listing of orion.employee_addresses
Employee_ Postal_
Employee_Name . . . State Country
ID Code
121044 Abbott, Ray . . . FL 33135 US
120145 Aisbitt, Sandy . . . 2001 AU
Akinfolarin,
120761 . . . PA 19145 US
Tameaka
120656 Amos, Salley . . . CA 92116 US
121107 Anger, Rose . . . PA 19142 US
121038 Anstey, David . . . FL 33157 US
120273 Antonini, Doris . . . FL 33141 US
. . . . . .
. . . . . .
. . . . . .

10
8-6 Chapter 8 Combining Data Horizontally

Business Scenario
You need to combine these two data sets to determine the
employee’s name and the employee’s manager’s name.
Partial Listing of names
Employee Manager_
Employee_Name Manager_Name
_ID ID
120102 120101 Zhou, Tom Lu, Patrick
120103 120101 Dawes, Wilson Lu, Patrick
120104 120101 Billington, Kareen Lu, Patrick
120105 120101 Povey, Liz Lu, Patrick
120121 120102 Elvish, Irenie Zhou, Tom
120122 120102 Ngan, Christina Zhou, Tom
120123 120102 Hotstone, Kimiko Zhou, Tom
120124 120102 Daymond, Lucian Zhou, Tom
. . . .
. . . .
. . . .

11

Steps for Merging

Step 1: Sort orion.employee_addresses by
Employee_ID to create the data set addresses.
Step 2: Merge addresses and orion.staff by
Employee_ID to create an intermediate data set.
Step 3: Sort the intermediate data set by Manager_ID.
Step 4: Merge the intermediate data set and addresses.
Use the RENAME= data set option to rename
Employee_ID to Manager_ID and
Employee_Name to Manager_Name
in the addresses data set.

12
 8.1 DATA Step Merges and SQL Procedure Joins 8-7

Using the DATA Step to Perform a Match-Merge

p308d01
proc sort data=orion.employee_addresses(keep=Employee_ID
Employee_Name)
out=addresses;
by Employee_ID;
run;

data temp1;
keep Employee_Name Employee_ID Manager_ID;
merge orion.staff(in=S keep=Employee_ID Manager_ID)
addresses(in=A);
by Employee_ID;
if S and A; /* Matches only */
run;

proc sort data=temp1;

by Manager_ID;
run;

data names;
merge temp1(in=T)
addresses(rename=(Employee_ID=Manager_ID
Employee_Name=Manager_Name) in=A);
by Manager_ID;
if A and T;
run;

proc print data=names(obs= 10);

title "Names Data Set";
run;
8-8 Chapter 8 Combining Data Horizontally

Advantages of a DATA Step Merge

Advantages of a DATA Step Merge
Multiple values can be returned.
There is no limit to the size of the tables, other than disk space.
Multiple BY variables enable lookups that depend on more than
one variable.
Multiple data sets can be used.
A merge enables complex business logic to be incorporated into
the new data set by using DATA step processing, such as arrays
and DO loops, in addition to merging features.
The IN= data set option and subsequent IF-THEN/ELSE logic
afford comprehensive control over whether to accept, reject, or
process an observation depending on which data set contributed to
the observation.
Observations with duplicate BY values are joined one-to-one
instead of being expanded into a Cartesian product as SQL does.
14

Disadvantages of a DATA Step Merge

Disadvantages of a DATA Step Merge
Data sets must be sorted or indexed based on the BY variable(s).
An exact match on the BY variable(s) value(s) must be found.
The BY variable(s) must be present in all data sets.
When more than one data set contributes variables with the same
name, the values from the variable in the rightmost data set
overwrite the other like-named variables, and no warning is printed.*

a b
* Example X Y X Y
1 2 1 3
data c;
merge a b; c
by X;
X Y
run;
1 3
15
 8.1 DATA Step Merges and SQL Procedure Joins 8-9

SQL Inner Join

When PROC SQL performs an equijoin, combinations of
observations from both data sets with matches on the
common variable remain.
proc sql;
create table three as
select one.*, two.Z
from one, two
where one.X=two.X; three
X Y Z
two 1 a f
one X Z
1 a r
X Y 1 f
1 a s
1 a 1 r 1 d f
1 d 1 s 1 d r
4 c 3 t 1 d s
5 g 4 w 4 c w
16

8.02 Multiple Choice Poll

By default, how does the SQL procedure perform an inner
join?
a. sequentially
b. creates a Cartesian product

18
8-10 Chapter 8 Combining Data Horizontally

Using a PROC SQL Join to Perform a Match-Merge

p308d02
proc sql;
create table namessql as
select e.Employee_ID,
e.Employee_Name,
Manager_ID,
m.Employee_Name as Manager_Name
from orion.staff,
orion.employee_addresses as e,
orion.employee_addresses as m
where e.Employee_ID=staff.Employee_ID
and m.Employee_ID=staff.Manager_ID
order by Manager_ID,
Employee_ID;
quit;

proc print data=namessql(obs=10);

title "Employee and Manager Names";
run;
 8.1 DATA Step Merges and SQL Procedure Joins 8-11

Advantages of PROC SQL Joins

Advantages of PROC SQL Joins
Multiple data sets can be joined without having common variables
in all data sets.
Data sets do not have to be sorted or indexed.

Inequality joins can be performed.

PROC SQL follows ANSI standard language definitions, so that you

can use the knowledge gained from other SQL implementations.

21

Disadvantages of PROC SQL Joins

Disadvantages of PROC SQL Joins
The maximum number of tables that can be joined at one time is
256.
For simple joins, PROC SQL might require more resources than
the DATA step with the MERGE statement.
Complex business logic is difficult to incorporate into the join.

Duplicate BY values are combined into a Cartesian product, which

can produce an extremely large output data set.

22
8-12 Chapter 8 Combining Data Horizontally

Comparing Merging and SQL

Match-Merge SQL Inner Join
There is no limit to the number of data sets nor The maximum number of tables that
the size of the data sets other than disk space. can be joined at one time is 256.

Data is processed sequentially so that Data is processed using a Cartesian

observations with duplicate BY values are product for duplicate BY values.
joined one-to-one.

Multiple data sets can be created. Only one data set can be created
with one CREATE TABLE statement.

Complex business logic can be incorporated CASE logic can be used for business
using IF-THEN or SELECT/WHEN logic. logic; however, it is not as flexible as
DATA step syntax.

The data sets being merged must be sorted or The data sets being joined do not
indexed on the BY variable(s). have to be sorted nor indexed.

An exact match on the BY-variable(s) value(s) Inequality joins can be performed.

must be found.

Like-named BY variables must be available in Common variables do not have to be

all data sets. in all data sets.
23

Comparison Programs
The DATA step merge and the PROC SQL inner join
do not always give you the same results.
The following programs are used to generate the results
for the next four result sets:
proc sql;
data three; create table three as
merge one two; select one.X, one.Y, two.Z
by X; from one, two
run; where one.X=two.X;
quit;

24
 8.1 DATA Step Merges and SQL Procedure Joins 8-13

Merge and SQL Join Comparison

ONE-TO-ONE matches produce identical results:
one two
X Y X Z
1 a 1 f
2 b 2 g

three – DATA step and PROC SQL

X Y Z
1 a f
2 b g

25

Merge and SQL Join Comparison

ONE-TO-MANY matches produce identical results:
one two
X Y X Z
1 a 1 f
2 b 1 r
2 g

three - DATA step and PROC SQL

X Y Z
1 a f
1 a r
2 b g
26
8-14 Chapter 8 Combining Data Horizontally

Merge and SQL Join Comparison

MANY-TO-MANY matches produce different results:
one two
X Y X Z
1 a 1 f
1 c 1 r
2 b 2 g

three - DATA step three - PROC SQL

X Y Z
X Y Z
1 a f
1 a f
1 a r
1 c r
1 c f
2 b g 1 c r
27 2 b g
 8.1 DATA Step Merges and SQL Procedure Joins 8-15

Reference Information

The following DATA step creates a Cartesian product:

data three(drop=Temp);
set one;
do i=1 to TotObs;
set two(rename=(X=Temp))
nobs=TotObs point=i;
if X=Temp then output;
end;
run;
In SAS 9.2, you can use the DATA step hash object with the MULTIDATA attribute to create a Cartesian
product.
data three;
drop rc;
length X 8 Y $1;
if _N_=1 then do;
declare hash H(dataset:'one', multidata:'yes');
H.definekey('X');
H.definedata('X', 'Y');
H.definedone();
end;
set two;
rc=H.find();
do while (rc=0);
output;
rc=H.find_next();
end;
run;
The MULTIDATA attribute specifies whether multiple data items are allowed for each key.
For more information, consult the SAS Help facility using the path shown below:
SAS Products Ö Base SAS Ö SAS 9.2 Language References: Concepts Ö DATA Step
Concepts Ö Using DATA Step Component Objects Ö Using the Hash Object
8-16 Chapter 8 Combining Data Horizontally

Merge and SQL Join Comparison

NONMATCHING data produces different results:
one two
X Y X Z
1 a 1 f
2 b 3 t
3 c 4 w

three - DATA step three - PROC SQL

X Y Z
X Y Z
1 a f
1 a f
3 c t
2 b
3 c t
4 w
28

Reference Information

The following SQL step produces results that are identical to those of the DATA step when there is
nonmatching data.
proc sql;
select coalesce(one.X, two.X) as X, Y, Z
from one full join two
on one.X=two.X;
quit;
The following DATA step merge produces results that are identical to those of the SQL inner join when
there is nonmatching data.
data three;
merge one(in=O) two(in=T);
by X;
if O and T;
run;
 8.1 DATA Step Merges and SQL Procedure Joins 8-17

Exercises (Optional)

Level 1

1. Merging or Joining Three Data Sets

The data set orion.order_fact has details about purchases made.
Partial Listing of orion.order_fact
Partial orion.order_fact

Order_ Delivery_
Obs Customer_ID Employee_ID Street_ID Date Date Order_ID

1 63 121039 9260125492 11JAN2003 11JAN2003 1230058123

2 5 99999999 9260114570 15JAN2003 19JAN2003 1230080101
3 45 99999999 9260104847 20JAN2003 22JAN2003 1230106883
4 41 120174 1600101527 28JAN2003 28JAN2003 1230147441
5 183 120134 1600100760 27FEB2003 27FEB2003 1230315085

Order_ Total_Retail_ CostPrice_

Obs Type Product_ID Quantity Price Per_Unit Discount

1 1 220101300017 1 $16.50 $7.45 .

2 2 230100500026 1 $247.50 $109.55 .
3 2 240600100080 1 $28.30 $8.55 .
4 1 240600100010 2 $32.00 $6.50 .
5 1 240200200039 3 $63.60 $8.80 .

The data set orion.customer_dim has the customers’ names.

Partial Listing of orion.customer_dim
Partial orion.customer_dim

Obs Customer_ID Customer_Name

1 4 James Kvarniq
2 5 Sandrina Stephano
3 9 Cornelia Krahl
4 10 Karen Ballinger
5 11 Elke Wallstab
8-18 Chapter 8 Combining Data Horizontally

The data set orion.product_dim has the product names and supplier names.
Partial Listing of orion.product_dim
Partial orion.product_dim

Obs Product_ID Product_Name Supplier_Name

1 210200100009 Kids Sweat Round Neck,Large Logo A Team Sports

2 210200100017 Sweatshirt Children's O-Neck A Team Sports
3 210200200022 Sunfit Slow Swimming Trunks Nautlius SportsWear Inc
4 210200200023 Sunfit Stockton Swimming Trunks Jr. Nautlius SportsWear Inc
5 210200300006 Fleece Cuff Pant Kid'S Eclipse Inc

a. Combine the three data sets to create a data set named purchases that contains the customer
name, product name, and supplier name for the customers in the orion.order_fact data set.
b. Order the data by Product_ID and print the first five observations of the purchases data set.
PROC PRINT Output
Partial purchases Data Set

Obs Customer_Name Product_Name Supplier_Name

1 Kyndal Hooks Kids Sweat Round Neck,Large Logo A Team Sports

2 Annmarie Leveille Sweatshirt Children's O-Neck A Team Sports
3 Najma Hicks Sunfit Slow Swimming Trunks Nautlius SportsWear Inc
4 Yan Kozlowski Sunfit Stockton Swimming Trunks Jr. Nautlius SportsWear Inc
5 Kyndal Hooks Fleece Cuff Pant Kid'S Eclipse Inc

Level 2

2. Merging or Joining Data to Create Multiple Data Sets

The data set orion.order_fact has details about purchases that were made.
Partial Listing of orion.order_fact
Partial orion.order_fact

Order_ Delivery_
Obs Customer_ID Employee_ID Street_ID Date Date Order_ID

1 63 121039 9260125492 11JAN2003 11JAN2003 1230058123

2 5 99999999 9260114570 15JAN2003 19JAN2003 1230080101
3 45 99999999 9260104847 20JAN2003 22JAN2003 1230106883
4 41 120174 1600101527 28JAN2003 28JAN2003 1230147441
5 183 120134 1600100760 27FEB2003 27FEB2003 1230315085

Order_ Total_Retail_ CostPrice_

Obs Type Product_ID Quantity Price Per_Unit Discount

1 1 220101300017 1 $16.50 $7.45 .

2 2 230100500026 1 $247.50 $109.55 .
3 2 240600100080 1 $28.30 $8.55 .
4 1 240600100010 2 $32.00 $6.50 .
5 1 240200200039 3 $63.60 $8.80 .
 8.1 DATA Step Merges and SQL Procedure Joins 8-19

The data set orion.customer_dim has the customers' names.

Partial Listing of orion.customer_dim
Partial orion.customer_dim

Obs Customer_ID Customer_Name

1 4 James Kvarniq
2 5 Sandrina Stephano
3 9 Cornelia Krahl
4 10 Karen Ballinger
5 11 Elke Wallstab

The data set orion.product_dim has the product names and supplier names.
Partial Listing of orion.product_dim
Partial orion.product_dim

Obs Product_ID Product_Name Supplier_Name

1 210200100009 Kids Sweat Round Neck,Large Logo A Team Sports

2 210200100017 Sweatshirt Children's O-Neck A Team Sports
3 210200200022 Sunfit Slow Swimming Trunks Nautlius SportsWear Inc
4 210200200023 Sunfit Stockton Swimming Trunks Jr. Nautlius SportsWear Inc
5 210200300006 Fleece Cuff Pant Kid'S Eclipse Inc

Combine the three data sets to create the following data sets:
• a data set named no_purchases that contains the customers who did not make any purchases
• a data set named purchases that contains the customer name, product name, and supplier name for
those customers in the orion.order_fact data set
• a data set named no_products that contains the product names and suppliers for products that were
not purchased
Partial Listing of no_purchases
no_purchases Data Set

Obs Customer_ID Customer_Name

1 33 Rolf Robak
2 42 Thomas Leitmann
8-20 Chapter 8 Combining Data Horizontally

Partial Listing of purchases

Partial purchases Data Set

Order_ Delivery_
Obs Customer_ID Employee_ID Street_ID Date Date Order_ID

1 90 121028 9260111614 19NOV2007 19NOV2007 1243960910

2 49 121035 9260104510 26NOV2004 26NOV2004 1234198497
3 79 99999999 9260101874 27MAY2005 03JUN2005 1235926178
4 52 121030 9260116235 12DEC2006 12DEC2006 1240886449
5 90 121032 9260111614 03MAY2007 03MAY2007 1242149082

Order_ Total_Retail_ CostPrice_

Obs Type Product_ID Quantity Price Per_Unit Discount Customer_Name

1 1 210200100009 2 $69.40 $15.50 . Kyndal Hooks

2 1 210200100017 1 $39.00 $17.35 . Annmarie Leveille
3 3 210200200022 2 $36.00 $7.05 . Najma Hicks
4 1 210200200023 1 $19.80 $8.25 . Yan Kozlowski
5 1 210200300006 1 $14.30 $7.70 . Kyndal Hooks

Obs Product_Name Supplier_Name

1 Kids Sweat Round Neck,Large Logo A Team Sports

2 Sweatshirt Children's O-Neck A Team Sports
3 Sunfit Slow Swimming Trunks Nautlius SportsWear Inc
4 Sunfit Stockton Swimming Trunks Jr. Nautlius SportsWear Inc
5 Fleece Cuff Pant Kid'S Eclipse Inc

Partial SAS Log

NOTE: There were 617 observations read from the data set WORK.TEMP.
NOTE: There were 481 observations read from the data set ORION.PRODUCT_DIM.
NOTE: The data set WORK.PURCHASES has 617 observations and 15 variables.
NOTE: The data set WORK.NO_PRODUCTS has 0 observations and 3 variables.

Level 3

3. Merging or Joining Multiple Data Sets

The data set orion.organization_dim has the levels of managers for each employee.

 Not all Manager_Leveln variables have values.

Partial Listing of orion.organization_dim

Partial orion.organization_dim

Manager_ Manager_ Manager_ Manager_ Manager_ Manager_

Obs Employee_ID Level1 Level2 Level3 Level4 Level5 Level6

1 120101 120261 120259 . . . .

2 120102 120101 120261 120259 . . .
3 120103 120101 120261 120259 . . .
4 120104 120101 120261 120259 . . .
5 120105 120101 120261 120259 . .
 8.1 DATA Step Merges and SQL Procedure Joins 8-21

The data set orion.employee_addresses contains the employee IDs and the employee names for all
employees.
Partial Listing of orion.employee_addresses
Partial orion.employee_addresses

Employee_
Obs ID Employee_Name

1 121044 Abbott, Ray

2 120145 Aisbitt, Sandy
3 120761 Akinfolarin, Tameaka
4 120656 Amos, Salley
5 121107 Anger, Rose

Create a data set named manager_names that contains the Employee_ID variable, the six
Manager_ID variables, and the six manager names.
Partial Listing of manager_names
Partial manager_names Data

Manager_ Manager_ Manager_ Manager_ Manager_ Manager_

Obs Employee_ID Level1 Level2 Level3 Level4 Level5 Level6 Manager1_Name

420 121144 121142 121141 120261 120259 . . Steiber, Reginald

421 121145 121142 121141 120261 120259 . . Steiber, Reginald
422 121146 121141 120261 120259 . . . Bleu, Henri Le
423 121147 121142 121141 120261 120259 . . Steiber, Reginald
424 121148 121141 120261 120259 . . . Bleu, Henri Le

Manager5_ Manager6_
Obs Manager2_Name Manager3_Name Manager4_Name Name Name

420 Bleu, Henri Le Highpoint, Harry Miller, Anthony

421 Bleu, Henri Le Highpoint, Harry Miller, Anthony
422 Highpoint, Harry Miller, Anthony
423 Bleu, Henri Le Highpoint, Harry Miller, Anthony
424 Highpoint, Harry Miller, Anthony
8-22 Chapter 8 Combining Data Horizontally

8.2 Using an Index to Combine Data

Objectives
„ Use the SET statement with the KEY= option to
combine two SAS data sets.
„ Use _IORC_ to determine whether the index search
was successful.

32

Business Scenario
The data set orion.catalog contains the order information
for catalog sales and has 38 observations.

Partial Listing of orion.catalog

Customer_ Total_Retail_
Order_ID Quantity
ID Price
5 1230080101 1 247.50
15 1240080101 3 216.50
45 1230106883 1 28.30
79 1230333319 1 234.60
23 1230338566 1 35.40
. . . .
. . . .
. . . .

33
 8.2 Using an Index to Combine Data 8-23

Business Scenario
The data set orion.customer_dim_more contains
information about customers and has 1,500 observations.
Partial Listing of orion.customer_dim_more
Customer_ Customer_ Customer Customer_ Customer_ Customer_ Customer_
ID Country _Gender Name ... Type Group Age
Orion Club
James Orion Club
4 US M
Kvarniq ... members low
members
33
activity
Orion Club Gold
Sandrina Orion Club
5 US F
Stephano ... members medium
Gold members
28
activity
Orion Club Gold
Cornelia Orion Club
9 DE F
Krahl ... members medium
Gold members
33
activity
Orion Club
Karen Orion Club
10 US F
Ballinger ... members high
members
23
activity
Orion Club
Elke Orion Club
11 DE F
Wallstab ... members high
members
33
activity
. . . . . . . .
. . . . . . . .
. . . . . . . .
34

Business Scenario
You need to combine the two data sets to create two new
data sets: one with information about the customers who
purchase products from the catalog for whom you have
demographics and the other for customers for whom you
do not have any information.
Partial PROC PRINT Output: catalog_customers PROC PRINT
Catalog Customers (Partial Output) Output: errors
Total_Retail_ No Demographic Data
Obs Customer_ID Order_ID Quantity Price Available
1 5 1230080101 1 $247.50
2 45 1230106883 1 $28.30 Obs Customer_ID
3 79 1230333319 1 $234.60
4 23 1230338566 1 $35.40 1 15
5 16 1230450371 2 $128.40 2 66
Customer_ Customer_ Customer_
Obs Country Gender Customer_Name Age

1 US F Sandrina Stephano 28
2 US F Dianne Patchin 28
3 US F Najma Hicks 21
4 US M Tulio Devereaux 58
5 DE M Ulrich Heyde 68

35
8-24 Chapter 8 Combining Data Horizontally

8.03 Multiple Choice Poll

If you use a DATA step MERGE to combine the data sets,
how many observations are read from the larger data set,
orion.customer_dim_more?
a. 38
b. 1,500

37

Combining a Large Data Set with a Small One

You can use multiple SET statements to combine the two
data sets.
data catalog_customers(keep=Customer_ID Order_ID Quantity
Total_Retail_Price
Customer_Country
Customer_Gender
Customer_Name
Customer_Age)
errors(keep=Customer_ID);
c set orion.catalog(keep=Customer_ID Order_ID
Quantity Total_Retail_Price);
d set orion.customer_dim_more key=Customer_ID;
if _IORC_=0 then output catalog_customers;
else do;
_ERROR_=0;
output errors;
end;
run;

p308d03
39

c The data set orion.catalog is read sequentially.

d The data set orion.customer_dim_more is read by direct access.
 8.2 Using an Index to Combine Data 8-25

Using the KEY= Option

An index is always used when a SET or MODIFY
statement contains the KEY= option.
Specify the KEY= option in the SET statement to use an
index to retrieve an observation that has key values equal
to the current value of the key variable(s).
General form of the KEY= option:

SET SAS-data-file-name KEY=index-name;

40

Assign a value to the index key variable(s) before the SET statement is executed. The index is then used
to retrieve an observation with the key value. WHERE processing is not enabled for a data set read with
the KEY= option.

Using the _IORC_ Automatic Variable

When you use the KEY= option, SAS creates an
automatic variable named _IORC_, which is an acronym
for input/output return code.
You can use the value of _IORC_ to determine whether
the search of the index was successful.

_IORC_=0 indicates that SAS found a matching

observation.
_IORC_ ne 0 indicates that the SET statement did not
successfully execute. One possible
cause is that SAS did not find a matching
observation.

41
8-26 Chapter 8 Combining Data Horizontally

Reference Information

Monitoring I/O Error Conditions

You can use the automatic variable _IORC_ with the %SYSRC AUTOCALL macro to test for specific
I/O error conditions that are created when you use the KEY= option in the SET statement.
General form for using %SYSRC with _IORC_:

IF _IORC_=%SYSRC(mnemonic) THEN…

Mnemonic Meaning

_DSENOM No matching observation was found.

_SOK The observation was located. _SOK has a value of 0.

The %SYSRC macro is in the AUTOCALL library. You must have the MACRO system option in effect
to use this macro. Consult SAS OnlineDoc for more information. Follow the path shown below:
Support & Training Ö Knowledge Base Ö Documentation Ö Base SAS Ö
SAS 9.2 Macro Language: Reference Ö Macro Language Dictionary Ö AutoCall Macros

Using the IORCMSG Function

The IORCMSG function returns the formatted error message that is associated with the current value of
the automatic variable _IORC_.
General form of the IORCMSG function:

character-variable=IORCMSG();

Character-variable specifies a character variable with a length of 200, unless the length was previously
assigned.
 8.2 Using an Index to Combine Data 8-27

Example:
p308d03a
data catalog_customers(keep=Customer_ID Order_ID Quantity
Total_Retail_Price
Customer_Country
Customer_Gender
Customer_Name
Customer_Age_Group)
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID
Quantity Total_Retail_Price);
set orion.customer_dim_more key=Customer_ID;
if _IORC_=0 then output catalog_customers;
else do;
output errors;
Message=iorcmsg();
_ERROR_=0;
putlog _N_ ' The problem is ' Message;
end;
run;

 The PUTLOG statement writes text to the log.

8-28 Chapter 8 Combining Data Horizontally

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
. . . . ... ... . 1
42
...

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
5 1230080101 1 247.50 ... ... . 1
43
...
 8.2 Using an Index to Combine Data 8-29

Execution Simplified Index on

orion.catalog (obs=2) orion.customer_dim_more
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
5 1230080101 1 247.50 ... ... . 1
44

Execution
Partial Listing of
orion.customer_dim_more
Simplified Index on
Customer Customer_ Customer Customer_
orion.customer_dim_more _ID Country _Gender Name
...
RID
Customer_ Record James
1 4 US M ...
Identifiers Kvarniq
ID
Sandrina
4 RID 2 5 US F ...
Stephano
5 RID Cornelia
3 9 DE F ...
Krahl
9 RID Karen
4 10 US F ...
. . Ballinger
. .
. . 5 Elke
11 DE F ...
Wallstab
13 RID
. . . .
16 RID . . . .
. . . . . .
. . Dianne
. . 45 US F ...
Patchin
45 RID . . . .
. . . . . .
. .
. . . .
. .

45
8-30 Chapter 8 Combining Data Horizontally

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
5 1230080101 1 247.50 ...
Stephano
... 0 1
46
...

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
True
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
5 1230080101 1 247.50 ...
Stephano
... 0 1
47
...
 8.2 Using an Index to Combine Data 8-31

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Output the current observation
Customer_Country
Customer_Gender
9 RID
. .
to catalog_customers.
Customer_Name
Customer_Age)
. .
. .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
end; Implicit RETURN; . .
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
5 1230080101 1 247.50 ...
Stephano
... 0 1
48
...

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
end; Initialize PDV. .
.
.
.
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
5 1230080101 1 247.50 ...
Stephano
... 0 2
49
...
8-32 Chapter 8 Combining Data Horizontally

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0;
output errors; The 2nd iteration 45
.
RID
.
end;
run; of the DATA step .
.
.
.
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 0 2
50
...

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2) No
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID
15
Identifiers
15 1240080101 3 216.50
4 RID
in
data catalog_customers(keep=Customer_ID Order_ID the
Quantity 5 RID
Total_Retail_Price index
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 1230015 2
51
...
 8.2 Using an Index to Combine Data 8-33

8.04 Quiz
Why do you not want this observation output to
catalog_customers?
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 1230015 2

53

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
False
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 1230015 2
55
...
8-34 Chapter 8 Combining Data Horizontally

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
Output current observation
set orion.catalog(keep=Customer_ID Order_ID 13 RID
Quantity
to errors.
Total_Retail_Price);
16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; Implicit RETURN; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 1230015 2
56
...

Execution Simplified Index on

orion.customer_dim_more
orion.catalog (obs=2)
Customer_ID Order_ID Quantity Total_Retail_Price Customer_ Record
5 1230080101 1 247.50 ID Identifiers
15 1240080101 3 216.50
4 RID
data catalog_customers(keep=Customer_ID Order_ID
Quantity 5 RID
Total_Retail_Price
Customer_Country 9 RID
Customer_Gender . .
Customer_Name . .
Customer_Age) . .
errors(keep=Customer_ID);
set orion.catalog(keep=Customer_ID Order_ID
Continue until EOF in 13 RID
Quantity
Total_Retail_Price);
orion.catalog. 16 RID
set orion.customer_dim_more key=Customer_ID; . .
if _IORC_=0 then output catalog_customers; . .
else do; . .
_ERROR_=0; 45 RID
output errors;
. .
end;
. .
run;
. .
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 1230015 2
57
...
 8.2 Using an Index to Combine Data 8-35

8.05 Quiz
Open and submit the program p308a01.
1. What messages do you see in your SAS log?
2. What is the value of _ERROR_?
3. Replace the ELSE statement with the following ELSE
DO group:
else do;
_ERROR_=0;
output errors;
end;
4. Resubmit the program and look at the log.
5. Why are there no messages now?

59

Using the _IORC_ Automatic Variable

To prevent writing the contents of the PDV to the log,
perform the following tasks:
„ Check the value of _IORC_.

„ Set _ERROR_ to 0, if there is no match.

63
8-36 Chapter 8 Combining Data Horizontally

Duplicate Key Values

data three;
set one;
set two key=Variable;
run;

Example 1: Contiguous duplications in one

one two
Variable Variable
A A
A A
A A

64 ...

If there are contiguous duplications in one, each of which has a match in two, then SAS performs a
one-to-one read.

Duplicate Key Values

data three;
set one;
set two key=Variable;
run;

Example 2: Contiguous duplications in one

one two
Variable Variable
A A
A A
A B No
Match
Run-time error
65 ...

If there are contiguous duplications in one, some of which do not have a match in two, then SAS
performs a one-to-one read until it finds a nonmatch. At that time, SAS encounters a run-time error.
 8.2 Using an Index to Combine Data 8-37

Duplicate Key Values

data three;
set one;
set two key=Variable/unique;
run;

Example 3: Contiguous duplications in one with the

UNIQUE option
one two
Variable Variable
A A
A A
A A

66

If there are contiguous duplications in one and the UNIQUE suboption in the KEY= option is used, then
SAS reads the first observation in two.

Duplicate Key Values

data three;
set one;
set two key=Variable;
run;

Example 4: Noncontiguous duplications in one

one two
Variable Variable
A A
B B
A A

 Using the UNIQUE option produces the same result.

67

If there are noncontiguous duplications in one, then SAS reads the first observation in two.
8-38 Chapter 8 Combining Data Horizontally

Using SET/SET with the KEY= Option

Advantages Disadvantages
Only the necessary An index on one data set is
observations are read. required.
An existing index is used. Creating and maintaining an index
uses resources.
_IORC_ can be used to When the indexed data set is not
control how nonmatching sorted by the key variable(s), there
data is handled. can be considerable increase in I/O.
This increase in I/O is because
of the random access of the data
set and the additional I/O required
to access the index.
The availability of DATA step
syntax provides the full
power of the DATA step.
68

Comparing SET/SET KEY=, Merging, and SQL

Match-Merge SQL Inner Join SET/SET KEY=
There is no limit to the The maximum number of Two or more data sets can
number of data sets nor tables that can be joined at be combined.
the size of the data sets one time is 256.
other than disk space.
Data is processed Data is processed using a The data set listed in the
sequentially so that Cartesian product for first SET statement is read
observations with duplicate BY values. sequentially; the data set in
duplicate BY values are the second and subsequent
joined one-to-one. SET statements is
processed via the index.
Observations in the first
data set with duplicates are
treated differently depending
on whether the duplicates
are contiguous or not.
Multiple data sets can be Only one data set can be Multiple data sets can be
created. created with one CREATE created.
TABLE statement.
69 continued...
 8.2 Using an Index to Combine Data 8-39

Comparing SET/SET KEY=, Merging, and SQL

Match-Merge SQL Inner Join SET/SET KEY=
Complex business logic The CASE clause can be Complex business logic can
can be incorporated using used for business logic. be incorporated using IF-
IF-THEN or However, it is not as THEN or SELECT/WHEN
SELECT/WHEN logic. flexible as DATA step logic.
syntax.

The data sets being The data sets being joined The data sets on all but the
merged must be sorted or do not have to be sorted first SET statement must
indexed on the BY nor indexed. have the index named on
variable(s). the KEY= option.

An exact match on the BY Inequality joins can be An exact match on the key
variable(s) value(s) must performed. value is required.
be found.

Like-named BY variables Common variables do not The indexed variable(s)

must be available in all have to be in all data sets. must be on all data sets.
data sets.

70
8-40 Chapter 8 Combining Data Horizontally

Using Multiple SET … KEY= Statements (Self-Study)

p308d04
You can use multiple SET statements with the KEY= option to combine several data sets. For example,
create a data set that contains the customers who ordered products from the Internet and from a catalog.
/***************************************************/
/* The index on the orion.customer_dim data set is */
/* created just for this example. It is deleted at */
/* the end of the program. */
/***************************************************/

proc sql;
create index Customer_ID
on orion.customer_dim(Customer_ID);
quit;

data catalog_internet others;

keep Customer_ID Order_ID Quantity
Total_Retail_Price Customer_Name
Int_OrderID Int_TotPrice Int_Quant
In_Dim In_Int In_Cat;

label Int_TotPrice='Total Retail Price for Internet Orders'

Int_Quant='Quantity of Internet Orders'
Total_Retail_Price='Total Retail Price for Catalog Orders'
Quantity='Quantity of Catalog Orders'
Order_ID='Catalog Order ID'
Int_OrderID='Internet Order ID'
In_Dim='In Customer_Dim data'
In_Int='In Internet data'
In_Cat='In Catalog data';

/* orion.catalog is read sequentially. InCat is created for */

/* educational purposes only. */

set orion.catalog(keep=Customer_ID Order_ID

Quantity Total_Retail_Price in=InCat);

/* orion.customer_dim is read using the index on Customer_ID */

/* InDim is created for educational purposes only. */

set orion.customer_dim(in=InDim) key=Customer_ID;

(Continued on the next page.)
 8.2 Using an Index to Combine Data 8-41

/* internet is read using the index on Customer_ID */

/* InInt is created for educational purposes only. */
/* Order_ID, Total_Retail_Price, Quantity are renamed */
/* so that both the values from orion.catalog data and the */
/* internet data are in the data set. Without the renaming, */
/* the values from internet would overwrite those from */
/* catalog. */

set orion.internet(in=InInt rename=(Order_ID=Int_OrderID

Total_Retail_Price=Int_TotPrice
Quantity=Int_Quant))
key=Customer_ID;

/* In_Dim, In_Int, and In_Cat are created for educational */

/* purposes only. They show which data set contributed */
/* to an observation. */

In_Dim=InDim;
In_Int=InInt;
In_Cat=InCat;

/* The value of _IORC_=0 only when data comes from both */

/* the orion.catalog and orion.internet data sets */
if _IORC_=0 then output catalog_internet;
else do;
_ERROR_=0;
output others;
end;
run;

proc sql;
drop index Customer_ID
from orion.customer_dim;
quit;

proc print data=catalog_internet label;

title 'Customers who order from both Catalog and Internet';
run;

proc print data=others label;

title 'Customers who ordered from the catalog only';
title2 'and the Customers who ordered from the catalog';
title3 'but were not in orion.Customer_Dim';
run;
8-42 Chapter 8 Combining Data Horizontally

Exercises

Level 1

4. Combining Data Sets Using an Index to Create One Data Set

The SAS data set orion.salesstaff contains information about the employees who work in Sales.
Partial Listing of orion.salesstaff
orion.salesstaff SAS Data Set
(Partial Output)

Birth_ Emp_Hire_
Employee_ID Job_Title Salary Gender Date Date

120121 Sales Rep. II $26,600 F 02AUG1944 01JAN1974

120134 Sales Rep. II $28,015 M 06JUN1949 01JAN1974
120151 Sales Rep. II $26,520 F 21NOV1944 01JAN1974
120154 Sales Rep. III $30,490 F 20JUL1944 01JAN1974
120166 Sales Rep. IV $30,660 M 14JUN1944 01JAN1974

Emp_Term_
Date Manager_ID SSN Employee_Name

. 120102 42-8321-982 Elvish, Irenie

30JUN2006 120102 905-76-7767 Shannan, Sian
. 120103 798-16-4924 Phaiyakounh, Julianna
. 120102 534-14-1428 Hayawardhana, Caterina
31AUG2006 120102 878-79-9390 Nowd, Fadi
 8.2 Using an Index to Combine Data 8-43

The SAS data set orion.organization_dim contains information about all employees. There is an
index on the Employee_ID variable.
Partial Listing of orion.organization_dim
orion.organization_dim SAS Data Set
(Partial Output)

Employee_
Employee_ID Country Company Department Section Org_Group

120101 AU Orion Australia Sales Management Sales Management Sales Management

120102 AU Orion Australia Sales Management Sales Management Sales Management
120103 AU Orion Australia Sales Management Sales Management Sales Management
120104 AU Orion Australia Administration Administration Administration
120105 AU Orion Australia Administration Administration Administration

Employee_ Employee_ Employee_

Job_Title Employee_Name Gender Salary BirthDate Hire_Date

Director Patrick Lu M $163,040 18AUG1976 01JUL2003

Sales Manager Tom Zhou M $108,255 11AUG1969 01JUN1989
Sales Manager Wilson Dawes M $87,975 22JAN1949 01JAN1974
Administration Manager Kareen Billington F $46,230 11MAY1954 01JAN1981
Secretary I Liz Povey F $27,110 21DEC1974 01MAY1999

Employee_ Manager_ Manager_ Manager_ Manager_ Manager_ Manager_ Manager_

Term_Date Levels Level1 Level2 Level3 Level4 Level5 Level6

. 2 120261 120259 . . . .
. 3 120101 120261 120259 . . .
. 3 120101 120261 120259 . . .
. 3 120101 120261 120259 . . .
. 3 120101 120261 120259 . . .

a. Create a SAS data set named sales_emps by using an index on Employee_ID to combine the
two data sets, orion.salesstaff and orion.organization_dim. Check the SAS log to ensure that
you do not have any data errors. Read only the variables Employee_ID, Department, Section,
and Org_Group from orion.organization_dim.
b. Print the first five observations of the sales_emps SAS data set.
PROC PRINT Output
Sales Employee Data
(Partial Output)

Birth_ Emp_Hire_ Emp_Term_

Obs Employee_ID Job_Title Salary Gender Date Date Date Manager_ID

1 120121 Sales Rep. II $26,600 F 02AUG1944 01JAN1974 . 120102

2 120134 Sales Rep. II $28,015 M 06JUN1949 01JAN1974 30JUN2006 120102
3 120151 Sales Rep. II $26,520 F 21NOV1944 01JAN1974 . 120103
4 120154 Sales Rep. III $30,490 F 20JUL1944 01JAN1974 . 120102
5 120166 Sales Rep. IV $30,660 M 14JUN1944 01JAN1974 31AUG2006 120102

Obs SSN Employee_Name Department Section Org_Group

1 42-8321-982 Elvish, Irenie Sales Sales Assorted Sports Articles

2 905-76-7767 Shannan, Sian Sales Sales Golf
3 798-16-4924 Phaiyakounh, Julianna Sales Sales Outdoors
4 534-14-1428 Hayawardhana, Caterina Sales Sales Racket Sports
5 878-79-9390 Nowd, Fadi Sales Sales Running Jogging
8-44 Chapter 8 Combining Data Horizontally

Level 2

5. Combining Data Sets Using an Index to Monitor Data Integrity

The data set orion.shoe_vendors contains information about the shoe products and their suppliers,
but there are vendors in orion.shoe_vendors whose products are not in the Orion Star price list,
which should contain all prices.
Partial Listing of orion.shoe_vendors
orion.shoe_vendors SAS Data Set
(Partial Output)

Product_ Product_ Product_

Line Category Group Product_ID Product_Name Supplier_ID

21 2102 2102004 210200400002 Deschutes Boys Outdoors Training Shoes 1303

21 2102 2102004 210200400005 Kid Air Terra Grande Running Shoes 1303
21 2102 2102004 210200400007 Kid Equivalent Street Shoes 1303
21 2102 2102004 210200400009 Kid Impeccably Strong(Bg) Basket Shoes 1303
21 2102 2102004 210200400012 Kid Trainer Lite V(Bp) Street Shoes 1303

Supplier_ Supplier_ Line_ Mfg_Suggested_

Name Country Group_Name Category_Name Name Retail_Price

Eclipse Inc US Eclipse, Kid's Shoes Children Sports Children $69.00

Eclipse Inc US Eclipse, Kid's Shoes Children Sports Children $73.00
Eclipse Inc US Eclipse, Kid's Shoes Children Sports Children $79.00
Eclipse Inc US Eclipse, Kid's Shoes Children Sports Children $114.00
Eclipse Inc US Eclipse, Kid's Shoes Children Sports Children $54.00

The data set orion.shoe_prices contains pricing information for all shoes.
Partial Listing of orion.shoe_prices
shoe_prices Data Set
(Partial Listing)

Total_Retail_ CostPrice_
Obs Product_ID Price Per_Unit

1 210200400002 $41.80 $21.00

2 210200400005 $97.80 $24.55
3 210200400007 $108.90 $18.25
4 210200400009 $56.50 $28.35
5 210200400012 $33.40 $16.80

Create a SAS data set named shoes and a SAS data set named errors by using an index on
Product_ID to combine the two data sets, orion.shoe_vendors and orion.shoe_prices.
 8.2 Using an Index to Combine Data 8-45

a. Create a simple index on the variable Product_ID in the data set orion.shoe_prices.
b. Read only the variables Product_ID, Product_Name, Supplier_Name, and
Mfg_Suggested_Retail_Price from orion.shoe_vendors.
Hint: There is a permanent format assigned to the Supplier_Country variable. To avoid a syntax
error, use the NOFMTERR system option.
c. Read only the variables Product_ID, Total_Retail_Price, CostPrice_Per_Unit from
orion.shoe_prices.
The shoes data set should have the price information for the shoe products.
d. The errors data set should contain data that is in orion.shoe_vendors, which is not in the
orion.shoe_prices data. The errors data set should contain only the variables Product_ID,
Product_Name, and Supplier_Name.

 The errors data set can then be used to determine why these vendors do not have
observations in price_list.
e. Delete the Product_ID index on the data set orion.shoe_prices.
f. Print the first five observations of the shoes SAS data set.
PROC PRINT Output
Shoe Data
(Partial Output)

Supplier_
Obs Product_ID Product_Name Name

1 210200400002 Deschutes Boys Outdoors Training Shoes Eclipse Inc

2 210200400005 Kid Air Terra Grande Running Shoes Eclipse Inc
3 210200400007 Kid Equivalent Street Shoes Eclipse Inc
4 210200400009 Kid Impeccably Strong(Bg) Basket Shoes Eclipse Inc
5 210200400012 Kid Trainer Lite V(Bp) Street Shoes Eclipse Inc

Mfg_Suggested_ Total_Retail_ CostPrice_

Obs Retail_Price Price Per_Unit

1 $69.00 $41.80 $21.00

2 $73.00 $97.80 $24.55
3 $79.00 $108.90 $18.25
4 $114.00 $56.50 $28.35
5 $54.00 $33.40 $16.80

 The data set shoes has 357 observations.

8-46 Chapter 8 Combining Data Horizontally

g. Print the observations of the errors SAS data set.

PROC PRINT Output
The errors Data

Supplier_
Obs Product_ID Product_Name Name

1 210200400027 Toddle Children's Air Mantra (3) (Bg) Shoes Eclipse Inc
2 210200400047 Toddler Fit Shoes Eclipse Inc
3 210201000174 Freestyle Children's Leather Street Shoes 3Top Sports
4 220200100123 Big Guy Men's Deschutz Slide Shoes Eclipse Inc

 The data set errors has four observations.

Level 3

6. Combining Data Sets Using an Index and Using the Macro Facility to Monitor Errors
The data set orion.first_internet_order contains the first order that a customer placed via the
Internet.
Partial Listing of orion.first_internet_order
orion.first_internet_order SAS Data Set
(Partial Output)

Order_ Delivery_
Customer_ID Employee_ID Street_ID Date Date Order_ID

4 99999999 9260106519 02MAR2004 03MAR2004 1232410925

5 99999999 9260114570 02MAY2007 07MAY2007 1242140006
9 99999999 3940106659 15APR2004 20APR2004 1232698281
11 99999999 3940108592 29OCT2003 03NOV2003 1231653765
19 99999999 3940106547 26DEC2003 30DEC2003 1231976710

Total_Retail_ CostPrice_
Product_ID Quantity Price Per_Unit Discount

240800200030 1 $47.70 $18.80 .

240100100159 1 $31.40 $13.90 .
230100600035 1 $29.40 $14.15 .
230100200047 1 $72.70 $35.20 .
240300100020 4 $56.40 $6.05 .

The data set orion.internet contains multiple orders that a customer placed via the Internet. There is
an index on the Order_ID variable.
 8.2 Using an Index to Combine Data 8-47

Partial Listing of orion.internet

orion.internet SAS Data Set
(Partial Output)

Order_ Delivery_
Customer_ID Employee_ID Street_ID Date Date Order_ID

70046 99999999 2600100017 02APR2003 03APR2003 1230500669

36 99999999 9260128237 18APR2003 20APR2003 1230591675
171 99999999 1600101555 01MAY2003 04MAY2003 1230657844
11171 99999999 2600100032 07MAY2003 09MAY2003 1230690733
17023 99999999 2600100021 20JUN2003 25JUN2003 1230931366

Total_Retail_ CostPrice_
Product_ID Quantity Price Per_Unit Discount

240200100131 2 $148.60 $41.35 .

240500100039 1 $34.50 $15.40 .
240100100646 1 $109.90 $46.80 .
240200100043 2 $282.40 $69.40 .
240200200007 2 $166.80 $8.35 .

a. Create a data set named processed_orders that contains the variables from
orion.first_internet_order and a variable named Comment. Use the index on the variable
Order_ID to retrieve the matching observation from orion.internet.
b. The variable Comment has the value Order has been processed if the Order_ID is in
both orion.first_internet_order and orion.internet. The value is Order has not been
processed if the Order_ID is not in both data sets.

c. Use the %SYSRC AUTOCALL macro described in the reference information in this chapter. In
addition, refer to SAS documentation by following the path shown below:
Support & Training Ö Knowledge Base Ö Documentation Ö Base SAS Ö
SAS 9.2 Macro Language: Reference Ö Macro Language Dictionary Ö AutoCall Macros
8-48 Chapter 8 Combining Data Horizontally

d. Print the first 10 observations of processed_orders.

PROC PRINT Output
Internet Orders
(Partial Output)

Order_ Delivery_
Obs Customer_ID Employee_ID Street_ID Date Date Order_ID Product_ID

1 4 99999999 9260106519 02MAR2004 03MAR2004 1232410925 240800200030

2 5 99999999 9260114570 02MAY2007 07MAY2007 1242140006 240100100159
3 9 99999999 3940106659 15APR2004 20APR2004 1232698281 230100600035
4 11 99999999 3940108592 29OCT2003 03NOV2003 1231653765 230100200047
5 19 99999999 3940106547 26DEC2003 30DEC2003 1231976710 240300100020
6 20 99999999 9260118934 18MAY2006 24MAY2006 1239226632 220200100190
7 24 99999999 9260115784 02JAN2007 05JAN2007 1241054779 240800200021
8 25 99999999 9260114570 15JAN2007 19JAN2007 1230080101 230100500026
9 27 99999999 9260105670 28JAN2007 02FEB2007 1241286432 240800200009
10 31 99999999 9260128428 25APR2007 29APR2007 1242076538 220200200022

Total_Retail_ CostPrice_
Obs Quantity Price Per_Unit Discount Comment

1 1 $47.70 $18.80 . Order has been processed.

2 1 $31.40 $13.90 . Order has been processed.
3 1 $29.40 $14.15 . Order has been processed.
4 1 $72.70 $35.20 . Order has been processed.
5 4 $56.40 $6.05 . Order has been processed.
6 3 $190.50 $29.95 . Order has been processed.
7 2 $195.60 $42.45 . Order has been processed.
8 1 $247.50 $109.55 . Order has not been processed.
9 2 $174.40 $34.90 . Order has been processed.
10 1 $57.30 $33.90 . Order has been processed.
 8.3 Combining Summary and Detail Data 8-49

8.3 Combining Summary and Detail Data

Objectives
„ Create an output SAS data set that contains
summary statistics from PROC SUMMARY.
„ Combine the output SAS data set from PROC
SUMMARY with a detail SAS data set.
„ Use the SQL procedure to combine summary and
detail data.
„ Use the SQL procedure to calculate the summary
statistic and combine it with every observation in the
data set.
„ Use the DATA step to calculate the summary
statistic and combine it with every observation in the
data set.

75

Business Scenario
The data set Partial Listing of
orion.totalsalaries has one orion.totalsalaries
observation for every value Manager_ Numemps DeptSal
ID
of Manager_ID.
120101 4 $269,570
Each observation contains 120102 48 $1,344,595
the number of people who 120103 30 $793,835
report to that manager, and 120104 15 $425,215
DeptSal is the total salary 120259 6 $941,155
for all of those employees. 120260 3 $216,065
120261 6 $595,935
120262 10 $545,255
120270 1 $43,635
120271 9 $280,155

76
8-50 Chapter 8 Combining Data Horizontally

Business Scenario
You need to calculate the total salaries paid by the
company. Then, divide each individual manager's
DeptSal by that total to create a variable named Percent.
Partial PROC PRINT Output
Percentage of Total Salaries
for Each Manager
(Partial Output)

Obs GrandTot Manager_ID Numemps DeptSal Percent

1 $15,695,800 120101 4 $269,570 1.72%

2 $15,695,800 120102 48 $1,344,595 8.57%
3 $15,695,800 120103 30 $793,835 5.06%
4 $15,695,800 120104 15 $425,215 2.71%
5 $15,695,800 120259 6 $941,155 6.00%
6 $15,695,800 120260 3 $216,065 1.38%
7 $15,695,800 120261 6 $595,935 3.80%
8 $15,695,800 120262 10 $545,255 3.47%
9 $15,695,800 120270 1 $43,635 0.28%
10 $15,695,800 120271 9 $280,155 1.78%

77

Combining Summary and Detail Data

The following is a common business task:
Step 1: Create a summary statistic from a data set
variable.
Step 2: Combine the summary information with detail
rows of the original data set.
Step 3: Calculate percentages.

78
 8.3 Combining Summary and Detail Data 8-51

Creating a Summary Data Set

The following techniques can be used to create
a summary data set:
„ the Output Delivery System (ODS)

„ the SUMMARY or MEANS procedure with

an OUTPUT statement
„ the DATA step

„ the SQL procedure

79

Reference Information

To use the Output Delivery System to calculate the sum statistic, use the following program:
p308d05
ods output summary=sumdata;

proc summary data=orion.totalsalaries print sum;

var DeptSal;
run;
The data set sumdata contains one observation with one variable, DeptSal_Sum, which has the value
15695800.
8-52 Chapter 8 Combining Data Horizontally

The SUMMARY Procedure

For numeric variables within a SAS data set, the
SUMMARY procedure computes descriptive statistics
such as the following:
„ mean

„ minimum

„ maximum

„ number of nonmissing values

„ standard deviation

80

Create the Summary Statistic

Step 1: Use the SUMMARY procedure to calculate the
total of the variable DeptSal.

proc summary data=orion.totalsalaries;

var DeptSal;
output out=summary sum=GrandTot;
run;

Listing of summary

Obs _TYPE_ _FREQ_ GrandTot

1 0 53 $15,695,800

p308d05
81

The output data set has variables that contain the requested statistics, plus the following variables:

_TYPE_ information about the class variables

_FREQ_ number of observations that an output level represents

 8.3 Combining Summary and Detail Data 8-53

PROC SUMMARY OUTPUT Statement

PROC SUMMARY can generate a report that contains the
descriptive statistics. To display the report, use the PRINT
option in the PROC SUMMARY statement.
The data produced by PROC SUMMARY is routed to
a SAS data set using an OUTPUT statement.
General form of the SUMMARY procedure:

PROC SUMMARY DATA=SAS-data-set <PRINT>;

OUTPUT OUT=SAS-data-set
output-statistic-specification(s);
RUN;

82

Combining Summary and Detail Data

Step 2: Combine the summary information with the detail
rows.
Step 3: Calculate the percentage.

data percent;
if _N_=1 then set summary(keep=GrandTot);
set orion.totalsalaries;
Percent=DeptSal / GrandTot;
format Percent percent8.2;
run;

p308d06
83

The _N_=1 condition causes the summary data set to be read only during the first iteration of the DATA
step. Without it, the DATA step reaches the end of file in summary on the second iteration of the DATA
step, and the DATA step terminates with one observation in the data set percent.
One observation from the data set orion.totalsalaries is read in each iteration of the DATA step.
8-54 Chapter 8 Combining Data Horizontally

Execution
summary True
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
. . . . . 1

84 ...

Execution
summary
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 . . . . 1

85 ...
 8.3 Combining Summary and Detail Data 8-55

Execution
summary
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 120101 4 269570 . 1

86 ...

Execution
summary
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 120101 4 269570 0.0172 1

87 ...
8-56 Chapter 8 Combining Data Horizontally

Execution
summary
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065 Implicit OUTPUT;
. . .
. . . Implicit RETURN;
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 120101 4 269570 0.0172 1

88 ...

Execution
summary
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065 Initialize PDV.
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 120101 4 269570 . 2

89 ...
 8.3 Combining Summary and Detail Data 8-57

Execution
summary False
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 120101 4 269570 . 2

90 ...

Execution
summary
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 120102 48 1344595 . 2

91 ...
8-58 Chapter 8 Combining Data Horizontally

Execution
Continue until EOF in
summary orion.totalsalaries.
GrandTot
15695800
data percent;
if _N_=1 then
Partial orion.totalsalaries set summary(keep=GrandTot);
Manager_ set orion.totalsalaries;
Numemps DeptSal
ID Percent=DeptSal/GrandTot;
120101 4 269570 format Percent percent8.2;
120102 48 1344595 run;
120103 30 793835
120104 15 425215
120259 6 941155
120260 3 216065
. . .
. . .
. . .

PDV
GrandTot Manager_ID Numemps DeptSal Percent D _N_
15695800 121145 45 1216055 0.077 53

94

8.06 Quiz
Open and submit the program p308a02.
1. How many observations are in the resulting data set?

2. Why?

3. How did you get 53 observations in the program

p308d06?

96
 8.3 Combining Summary and Detail Data 8-59

Combining Data Using the SQL Procedure

You can join a summary data set and a detail data set
using the SQL procedure.
proc sql;
create table percentsql as
select Manager_ID,
DeptSal,
GrandTot,
DeptSal / GrandTot as Percent
format=percent8.2
from orion.totalsalaries,
summary;
quit;

This program takes advantage of the default Cartesian

product that SQL creates with inner joins.
p308d07
100
8-60 Chapter 8 Combining Data Horizontally

Combining Data Using the SQL Procedure

You can also remerge overall summary results, such
as grand totals, with detail data using SQL.
proc sql;
create table percentsql as
select Manager_ID,
DeptSal,
sum(DeptSal) as GrandTot,
DeptSal / calculated GrandTot
as Percent format=8.2
from orion.totalsalaries;
quit;

p308d08
101

The SUM function with one argument calculates the total for the column DeptSal.
Because the alias GrandTot is assigned to the sum(DeptSal) column, the SELECT statement can use the
CALCULATED keyword to refer to GrandTot as the denominator in this calculation.
When SQL remerges summary data, it puts a note in the SAS log.
SAS Log
proc sql;
2 create table percentsql as
3 select Manager_ID,
4 DeptSal,
5 sum(DeptSal) as GrandTot,
6 DeptSal/calculated GrandTot
7 as Percent format=8.2
8 from orion.totalsalaries;
NOTE: The query requires remerging summary statistics back with the original data.
NOTE: Table WORK.PERCENTSQL created, with 53 rows and 4 columns.

9 quit;
NOTE: PROCEDURE SQL used (Total process time):
real time 0.39 seconds
 8.3 Combining Summary and Detail Data 8-61

In addition to using the SQL procedure for calculating the percentages in one step, the REPORT
procedure and the TABULATE procedure can calculate percentages in one step.
p308d08a
proc report data=orion.totalsalaries
out=report_pct(drop=_break_)
nowd;
column Manager_ID DeptSal DeptSal=PctSal;
define Manager_ID / display 'Manager ID';
define DeptSal / sum 'Department Salaries';
define PctSal / pctsum format=percent8.2
'Percent of Total Salaries';
run;

proc tabulate data=orion.totalsalaries

out=tab_pct(drop=_type_ _page_ _table_);
class Manager_ID;
var DeptSal;
table Manager_ID='Manager ID',
DeptSal='Department Salaries' * (sum*f=dollar14.2 pctsum);
run;
8-62 Chapter 8 Combining Data Horizontally

Using the DATA Step

You can use the DATA step to read the data from
orion.totalsalaries, calculate the GrandTot variable, and
then reread the data in order to calculate the percentages.

data percent(drop=i);
c if _N_=1 then do i=1 to TotObs;
d set orion.totalsalaries(keep=DeptSal)
nobs=TotObs;
e GrandTot + DeptSal;
end;
f set orion.totalsalaries;
g Percent=DeptSal / GrandTot;
format Percent percent8.2;
run;

p308d09
102

c During the first execution of the DATA step, the DO loop executes the SET statement for each
observation in the orion.totalsalaries data set.
d When the SET statement executes, it reads the value of DeptSal from orion.totalsalaries.
e The SUM statement GrandTot + DeptSal accumulates the value of DeptSal into the variable
GrandTot.
f The DO loop completes execution when i is greater than TotObs, preventing SAS from reaching the
end-of-file marker. The second SET statement reads the observations from orion.totalsalaries starting
with observation 1.
g The variable Percent is calculated for each of those observations.
 8.3 Combining Summary and Detail Data 8-63

Reference Information

In SAS 9.2 you can use the SUM method for the hash object to calculate the grand total of the variable
DeptSal.
p308d10
data tot_sal / view=tot_sal;
set orion.totalsalaries;
Key='A';
run;

data percent;
retain GrandTot 0;
if _N_=1 then do;
dcl hash H(suminc:'DeptSal');
H.definekey('Key');
H.definedone();
do while(not Done);
set tot_sal end=Done;
H.ref();
end;
H.sum(sum:GrandTot);
end;
set orion.totalsalaries;
Percent=DeptSal / GrandTot;
format Percent percent8.2;
run;
8-64 Chapter 8 Combining Data Horizontally

Exercises

Level 1

7. Combining Summary Data Containing an Average with Detail Data

The data set orion.customer_dim contains the age for each customer.
Partial Listing of orion.customer_dim
orion.customer_dim SAS Data Set
(Partial Output)

Customer_ Customer_ Customer_ Customer_ Customer_ Customer_

Customer_ID Country Gender Customer_Name FirstName LastName BirthDate Age_Group

4 US M James Kvarniq James Kvarniq 27JUN1974 31-45 years

5 US F Sandrina Stephano Sandrina Stephano 09JUL1979 15-30 years
9 DE F Cornelia Krahl Cornelia Krahl 27FEB1974 31-45 years
10 US F Karen Ballinger Karen Ballinger 18OCT1984 15-30 years
11 DE F Elke Wallstab Elke Wallstab 16AUG1974 31-45 years

Customer_
Customer_Type Customer_Group Age

Orion Club members low activity Orion Club members 33

Orion Club Gold members medium activity Orion Club Gold members 28
Orion Club Gold members medium activity Orion Club Gold members 33
Orion Club members high activity Orion Club members 23
Orion Club members high activity Orion Club members 33

a. Calculate the average age of all customers.

b. Create a SAS data set named age_dif, which combines the average age of all customers with the
orion.customer_dim data set in order to determine the difference between each customer's age
and the average for all customers. (You can use any method presented in this section.)
c. Print the first five observations of the age_dif SAS data set.
PROC PRINT Output
The age_dif Data Set
(Partial Output)

Customer_
Obs AvgAge Customer_ID Age Age_Difference

1 41.9740 4 33 -8.9740
2 41.9740 5 28 -13.9740
3 41.9740 9 33 -8.9740
4 41.9740 10 23 -18.9740
5 41.9740 11 33 -8.9740
 8.3 Combining Summary and Detail Data 8-65

Level 2

8. Combining Summary Data Containing a Total with Detail Data

The data set orion.employee_donations contains the contributions of employees to various
charities. The contributions are reported for each of four quarters.
Partial Listing of orion.employee_donations
orion.employee_donations SAS Data Set
(Partial Output)

Employee_ID Qtr1 Qtr2 Qtr3 Qtr4 Recipients

120265 . . . 25 Mitleid International 90%, Save the Baby Animals 10%

120267 15 15 15 15 Disaster Assist, Inc. 80%, Cancer Cures, Inc. 20%
120269 20 20 20 20 Cancer Cures, Inc. 10%, Cuidadores Ltd. 90%
120270 20 10 5 . AquaMissions International 10%, Child Survivors 90%
120271 20 20 20 20 Cuidadores Ltd. 80%, Mitleid International 20%

Paid_By

Cash or Check
Payroll Deduction
Payroll Deduction
Cash or Check
Payroll Deduction

a. Select any method to create a SAS data set named compare by performing the following tasks:
• Calculate the total contribution for each employee.
• Determine the average of the total contribution for all of the employees.
• Calculate the difference between the average and each individual employee's total
contribution.
b. Print the first five observations of the compare SAS data set.
PROC PRINT Output
The compare Data Set
(Partial Output)

Avg_
Obs Donation Employee_ID Qtr1 Qtr2 Qtr3 Qtr4

1 47.2581 120265 . . . 25
2 47.2581 120267 15 15 15 15
3 47.2581 120269 20 20 20 20
4 47.2581 120270 20 10 5 .
5 47.2581 120271 20 20 20 20

Total_
Obs Recipients Paid_By Donation Difference

1 Mitleid International 90%, Save the Baby Animals 10% Cash or Check 25 -22.2581
2 Disaster Assist, Inc. 80%, Cancer Cures, Inc. 20% Payroll Deduction 60 12.7419
3 Cancer Cures, Inc. 10%, Cuidadores Ltd. 90% Payroll Deduction 80 32.7419
4 AquaMissions International 10%, Child Survivors 90% Cash or Check 35 -12.2581
5 Cuidadores Ltd. 80%, Mitleid International 20% Payroll Deduction 80 32.7419
8-66 Chapter 8 Combining Data Horizontally

Level 3

9. Combining Summary Data Containing a Weighted Average and Detail Data

The SAS data set orion.order_fact contains the variables CostPrice_Per_Unit and Quantity.
Partial Listing of orion.order_fact
orion.order_fact SAS Data Set
(Partial Output)

Order_ Delivery_
Customer_ID Employee_ID Street_ID Date Date Order_ID

63 121039 9260125492 11JAN2003 11JAN2003 1230058123

5 99999999 9260114570 15JAN2003 19JAN2003 1230080101
45 99999999 9260104847 20JAN2003 22JAN2003 1230106883
41 120174 1600101527 28JAN2003 28JAN2003 1230147441
183 120134 1600100760 27FEB2003 27FEB2003 1230315085

Order_ Total_Retail_ CostPrice_

Type Product_ID Quantity Price Per_Unit Discount

1 220101300017 1 $16.50 $7.45 .

2 230100500026 1 $247.50 $109.55 .
2 240600100080 1 $28.30 $8.55 .
1 240600100010 2 $32.00 $6.50 .
1 240200200039 3 $63.60 $8.80 .

The data set orion.product_dim contains the variables Product_ID and Product_Name.
Partial Listing of orion.product_dim
orion.product_dim SAS Data Set
(Partial Output)

Product_ Product_
Product_ID Line Category Product_Group Product_Name

210200100009 Children Children Sports A-Team, Kids Kids Sweat Round Neck,Large Logo
210200100017 Children Children Sports A-Team, Kids Sweatshirt Children's O-Neck
210200200022 Children Children Sports Bathing Suits, Kids Sunfit Slow Swimming Trunks
210200200023 Children Children Sports Bathing Suits, Kids Sunfit Stockton Swimming Trunks Jr.
210200300006 Children Children Sports Eclipse, Kid's Clothes Fleece Cuff Pant Kid'S

Supplier_
Country Supplier_Name Supplier_ID

US A Team Sports 3298

US A Team Sports 3298
US Nautlius SportsWear Inc 6153
US Nautlius SportsWear Inc 6153
US Eclipse Inc 1303
 8.3 Combining Summary and Detail Data 8-67

a. Select any method to create a SAS data set named products by performing the following tasks:
• Calculate the total CostPrice_Per_Unit weighted by Quantity.
• Combine the weighted total with the orion.order_fact data. Create a new variable named
Percent that is based on the actual total cost (CostPrice_Per_Unit *Quantity) and the
weighted total.
b. Print the first five observations of the products SAS data set.
PROC PRINT Output
The products Data Set
(Partial Output)

CostPrice_
Obs Customer_ID Quantity Per_Unit Product_Name Percent

1 90 2 $15.50 Kids Sweat Round Neck,Large Logo 0.068%

2 49 1 $17.35 Sweatshirt Children's O-Neck 0.038%
3 79 2 $7.05 Sunfit Slow Swimming Trunks 0.031%
4 52 1 $8.25 Sunfit Stockton Swimming Trunks Jr. 0.018%
5 90 1 $7.70 Fleece Cuff Pant Kid'S 0.017%
8-68 Chapter 8 Combining Data Horizontally

Reference Information

To create a running total for a variable, you can use either the DATA step or the SQL procedure.
p308d11
proc sort data=orion.order_fact out=order_fact;
by Order_Date Order_ID;
run;

data running_totals;
keep Order_Date Product_ID Total_Retail_Price
Sum_Total_Retail_Price;
set order_fact;
Sum_Total_Retail_Price + Total_Retail_Price;
format Sum_Total_Retail_Price dollar8.2;
run;

proc print data=running_totals(obs=10);

run;

Running Totals using the DATA Step

Sum_Total_
Order_ Total_Retail_ Retail_
Obs Date Product_ID Price Price

1 11JAN2003 220101300017 $16.50 $16.50

2 15JAN2003 230100500026 $247.50 $264.00
3 20JAN2003 240600100080 $28.30 $292.30
4 28JAN2003 240600100010 $32.00 $324.30
5 27FEB2003 240200200039 $63.60 $387.90
6 02MAR2003 240100400005 $234.60 $622.50
7 03MAR2003 240800200062 $35.40 $657.90
8 03MAR2003 240800200063 $73.80 $731.70
9 09MAR2003 240500100004 $127.00 $858.70
10 09MAR2003 240500200003 $23.20 $881.90
 8.3 Combining Summary and Detail Data 8-69

p308d11
proc sql;
create table order_fact_with_obsnum as
select monotonic() as obsnum,
*
from orion.order_fact;
create table running_totals_sql as
select o1.Order_Date,
o1.Product_ID,
o1.Total_Retail_Price,
(select sum(o2.Total_Retail_Price)
from order_fact_with_obsnum as o2
where o2.obsnum <= o1.obsnum) as Sum_Total_Retail_Price
format=dollar8.2
from order_fact_with_obsnum as o1
order by Order_Date, Order_ID, Sum_Total_Retail_Price;

title 'Running Totals using PROC SQL;

select * from running_totals_sql(obs=10);
quit;

 The monotonic function enables you to create row numbers in SQL that are written to the table
and not only a displayed value. This Base SAS function returns 1 the first time that it is called, 2
the second time, 3 the next time, and so forth. See SAS Usage Note 15138 for more information
about the monotonic function.
Running Totals using PROC SQL

Date
Order was Total Retail
placed by Price for Sum_Total_
Customer Product ID This Product Retail_Price
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
11JAN2003 220101300017 $16.50 $16.50
15JAN2003 230100500026 $247.50 $264.00
20JAN2003 240600100080 $28.30 $292.30
28JAN2003 240600100010 $32.00 $324.30
27FEB2003 240200200039 $63.60 $387.90
02MAR2003 240100400005 $234.60 $622.50
03MAR2003 240800200062 $35.40 $657.90
03MAR2003 240800200063 $73.80 $731.70
09MAR2003 240500100004 $127.00 $858.70
09MAR2003 240500200003 $23.20 $881.90
8-70 Chapter 8 Combining Data Horizontally

To ensure that the data sets are the same, you can use PROC COMPARE.
p308d11
proc compare data=running_totals compare=running_totals_sql;
title 'Comparing the Resulting Data Sets';
run;

Comparing the Resulting Data Sets

The COMPARE Procedure

Comparison of WORK.RUNNING_TOTALS with WORK.RUNNING_TOTALS_SQL
(Method=EXACT)

Data Set Summary

Dataset Created Modified NVar NObs

WORK.RUNNING_TOTALS 28MAY08:14:40:19 28MAY08:14:40:19 4 617

WORK.RUNNING_TOTALS_SQL 28MAY08:14:44:21 28MAY08:14:44:21 4 617

Variables Summary

Number of Variables in Common: 4.

Observation Summary

Observation Base Compare

First Obs 1 1
Last Obs 617 617

Number of Observations in Common: 617.

Total Number of Observations Read from WORK.RUNNING_TOTALS: 617.
Total Number of Observations Read from WORK.RUNNING_TOTALS_SQL: 617.

Number of Observations with Some Compared Variables Unequal: 0.

Number of Observations with All Compared Variables Equal: 617.

NOTE: No unequal values were found. All values compared are exactly equal.
 8.4 Combining Data Conditionally (Self-Study) 8-71

8.4 Combining Data Conditionally (Self-Study)

Objectives
„ Combine data conditionally using multiple SET
statements.
„ Combine data conditionally with the SQL procedure.
„ Combine data conditionally using a hash object.

106

Business Scenario
Some combinations of data are based on a condition.
The data set orion.order_fact contains the
Total_Retail_Price for all values of Order_Date.
orion.order_fact(where=(Order_Date between
'01SEP2007'd and '30SEP2007'd))
Customer Employee Total_Retail CostPrice
Street_ID Order_Date . . . Discount
_ID _ID _Price _Per_Unit
928 99999999 9050100016 04SEP2007 ... $86.30 $41.40 .
27 99999999 9260105670 05SEP2007 ... $78.40 $16.45 .
31 121057 9260128428 06SEP2007 ... $50.30 $25.25 .
45 121065 9260104847 06SEP2007 ... $78.20 $39.20 .
5 121026 9260114570 09SEP2007 ... $52.50 $22.25 .
12 121051 9260103713 18SEP2007 ... $87.20 $44.95 .
69 121029 9260116402 20SEP2007 ... $23.50 $9.20 .
24 99999999 9260115784 25SEP2007 ... $46.10 $19.70 .
41 120195 1600101527 26SEP2007 ... $134.00 $28.90 .
11 99999999 3940108592 28SEP2007 ... $78.20 $19.65 .

 orion.order_fact is sorted by Order_Date.

107
8-72 Chapter 8 Combining Data Horizontally

Business Scenario
The data set orion.rates has the average conversion
rate for converting from dollars to euros for the weeks in
September 2007.
orion.rates
SDate EDate AvgRate
01SEP2007 07SEP2007 0.73117
08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725

 orion.rates is sorted by SDate.

108

Business Scenario
You need to determine the Total_Retail_Price in euros.
Listing of euros
Total_
Customer_ Order_ Product_
Retail_ SDate EDate AvgRate EuroPrice
ID Date ID
Price
928 04SEP2007 230100600030 $86.30 01SEP2007 07SEP2007 0.73117 € 63.10
27 05SEP2007 240500200082 $78.40 01SEP2007 07SEP2007 0.73117 € 57.32
31 06SEP2007 220200100137 $50.30 01SEP2007 07SEP2007 0.73117 € 36.78
45 06SEP2007 230100600015 $78.20 01SEP2007 07SEP2007 0.73117 € 57.18
5 09SEP2007 210200500016 $52.50 08SEP2007 14SEP2007 0.72184 € 37.90
12 18SEP2007 240200100053 $87.20 15SEP2007 21SEP2007 0.71589 € 62.43
69 20SEP2007 210200700016 $23.50 15SEP2007 21SEP2007 0.71589 € 16.82
24 25SEP2007 240600100102 $46.10 22SEP2007 30SEP2007 0.70725 € 32.60
41 26SEP2007 210200600067 $134.00 22SEP2007 30SEP2007 0.70725 € 94.77
11 30SEP2007 220200100002 $78.20 28SEP2007 30SEP2007 0.70725 € 55.31

109
 8.4 Combining Data Conditionally (Self-Study) 8-73

Conditionally Combining Data

What needs to be done:

Partial PDV
Order_Date SDate EDate AvgRate
04SEP2007 01SEP2007 07SEP2007 0.73117

Order_Date between
SDate and EDate

110 ...

Conditionally Combining Data

What needs to be done:

Partial PDV
Order_Date SDate EDate AvgRate
04SEP2007 01SEP2007 07SEP2007 0.73117

Order_Date between True Use

SDate and EDate this
rate.

112 ...
8-74 Chapter 8 Combining Data Horizontally

Conditionally Combining Data

What needs to be done:

Partial PDV
Order_Date SDate EDate AvgRate
09SEP2007 01SEP2007 07SEP2007 0.73117

Order_Date between False

SDate and EDate

114 ...

Conditionally Combining Data

What needs to be done:

Partial PDV
Order_Date SDate EDate AvgRate
09SEP2007 08SEP2007 14SEP2007 0.72184

Read observations from

orion.rates until
order_date is between
SDate and Edate.

115
 8.4 Combining Data Conditionally (Self-Study) 8-75

Conditionally Combining Data

What needs to be done:

Order_Date between SDate and EDate

True False

Use the rate Read a

in the PDV. new rate.

116

8.07 Poll
Can the DATA step merge be used for this task?
€ Yes
€ No

118
8-76 Chapter 8 Combining Data Horizontally

Conditionally Combining Data

The MERGE statement can be used to join data when
one of the following conditions is met:
„ The data can be joined data three;
by comparing values merge one two;
of a common BY value. by X;
run;

„ The data can be

combined by observation
number. In this case, data three;
merge one two;
there is no BY statement
run;
in the DATA step.

120

Multiple SET Statements (Review)

You can use multiple SET statements to combine
observations from several SAS data sets.
When you use multiple SET statements, the following
occurs:
„ Processing stops when SAS encounters the end-of-file
marker on either data set.
„ The variables in the PDV are not reinitialized when a
second SET statement is executed.
data euros;
set orion.order_fact(where=(Order_Date
between '01SEP2007'd
and '30SEP2007'd));
set orion.rates;
run;
121
 8.4 Combining Data Conditionally (Self-Study) 8-77

Conditionally Combining Data

data euros;
set orion.order_fact(where=(Order_Date between
'01SEP2007'd and '30SEP2007'd)
keep=Customer_ID Order_Date
Product_ID
Total_Retail_Price);
do while (not (SDate le Order_Date le EDate));
set orion.rates;
end;
EuroPrice=Total_Retail_Price * AvgRate;
format EuroPrice Euro10.2;
run;

orion.order_fact must be sorted by Order_Date.

orion.rates must be sorted by SDate.
p308d12
122 ...

To use multiple SET statements in this fashion, both data sets must be sorted in order (ascending or
descending) by the variables tested in the DO WHILE statement.

Conditionally Combining Data

data euros;
When this is true,
set orion.order_fact(where=(Order_Date between
you do not need to
'01SEP2007'd and '30SEP2007'd)
read a new rate.
keep=Customer_ID Order_Date
Product_ID
Total_Retail_Price);
do while (not (SDate le Order_Date le EDate));
set orion.rates;
end;
EuroPrice=Total_Retail_Price * AvgRate;
format EuroPrice Euro10.2;
run;

orion.order_fact must be sorted by Order_Date.

orion.rates must be sorted by SDate.
p308d12
123 ...
8-78 Chapter 8 Combining Data Horizontally

Conditionally Combining Data

data euros;
When this is true,
set orion.order_fact(where=(Order_Date between
you do need to
'01SEP2007'd and '30SEP2007'd)
read a new rate.
keep=Customer_ID Order_Date
Product_ID
Total_Retail_Price);
do while (not (SDate le Order_Date le EDate));
set orion.rates;
end;
EuroPrice=Total_Retail_Price * AvgRate;
format EuroPrice Euro10.2;
run;

orion.order_fact must be sorted by Order_Date.

orion.rates must be sorted by SDate.
p308d12
124

8.08 Quiz
Why do you have to use a WHERE= data set option
rather than a WHERE statement to subset by
Order_Date?

127
 8.4 Combining Data Conditionally (Self-Study) 8-79

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 . . . . 1

129 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
False
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 . . . . 1

Is 04SEP2007 between . and .?

130 ...
8-80 Chapter 8 Combining Data Horizontally

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
run;
SDate EDate AvgRate The DO WHILE condition
01SEP2007 07SEP2007 0.73117 is true, so the DO loop
08SEP2007 14SEP2007 0.72184 executes.
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
True
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 . . . . 1

Is NOT (04SEP2007 between . and .)?

131 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 01SEP2007 07SEP2007 0.73117 . 1

132 ...
 8.4 Combining Data Conditionally (Self-Study) 8-81

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;
The DO WHILE condition
01SEP2007 07SEP2007 0.73117 is false, so the DO loop
08SEP2007 14SEP2007 0.72184
does not execute.
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ False
Retail_ SDate EDate
Avg Euro D
_N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 01SEP2007 07SEP2007 0.73117 . 1

Is NOT( 04SEP2007 between 01SEP2007 and 07SEP2007)?

135 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 01SEP2007 07SEP2007 0.73117 63.10 1

136 ...
8-82 Chapter 8 Combining Data Horizontally

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 Implicit
07SEP2007OUTPUT;
0.73117
08SEP2007
Implicit RETURN;
14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 01SEP2007 07SEP2007 0.73117 63.10 1

137

8.09 Quiz
What variables are set to missing at the top of the DATA
step?

139
 8.4 Combining Data Conditionally (Self-Study) 8-83

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . Initialize
. PDV. set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
928 04SEP2007 230100600030 86.30 01SEP2007 07SEP2007 0.73117 . 2

141 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
27 05SEP2007 240500200082 78.40 01SEP2007 07SEP2007 0.73117 . 2

142 ...
8-84 Chapter 8 Combining Data Horizontally

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;
The DO WHILE condition
01SEP2007 07SEP2007 0.73117 is false, so the DO loop
08SEP2007 14SEP2007 0.72184
does not execute.
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Customer Order_ Product_ False
Total_
Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
27 05SEP2007 240500200082 78.40 01SEP2007 07SEP2007 0.73117 . 2

Is NOT(05SEP2007 between 01SEP2007 and 07SEP2007)?

144 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
27 05SEP2007 240500200082 78.40 01SEP2007 07SEP2007 0.73117 57.32 2

145 ...
 8.4 Combining Data Conditionally (Self-Study) 8-85

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate Continue until
run;

01SEP2007 07SEP2007 0.73117 Order_Date

08SEP2007 14SEP2007 0.72184
is 09SEP2007.
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
5 09SEP2007 210200500016 52.50 01SEP2007 07SEP2007 0.73117 . 5

147 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ True
Retail_ SDate EDate
Avg Euro D
_N_
_ID Date ID Rate Price
Price
5 09SEP2007 210200500016 52.50 01SEP2007 07SEP2007 0.73117 . 5

Is NOT(09SEP2007 between 01SEP2007 and 07SEP2007)?

149 ...
8-86 Chapter 8 Combining Data Horizontally

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

The DO WHILE condition
08SEP2007 14SEP2007 0.72184
is true, so the DO loop
15SEP2007 21SEP2007 0.71589 executes.
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ True
Retail_ SDate EDate
Avg Euro D
_N_
_ID Date ID Rate Price
Price
5 09SEP2007 210200500016 52.50 08SEP2007 14SEP2007 0.72184 . 5

150 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;
The DO WHILE condition
01SEP2007 07SEP2007 0.73117 is false, so the DO loop
08SEP2007 14SEP2007 0.72184
does not execute.
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ False
Retail_ SDate EDate
Avg Euro D
_N_
_ID Date ID Rate Price
Price
5 09SEP2007 210200500016 52.50 08SEP2007 14SEP2007 0.72184 . 5

Is NOT(09SEP2007 between 08SEP2007 and 14SEP2007)?

152 ...
 8.4 Combining Data Conditionally (Self-Study) 8-87

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. . . . . .
end;
orion.rates EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
5 09SEP2007 210200500016 52.50 08SEP2007 14SEP2007 0.72184 37.90 5

153 ...

Execution
Partial orion.order_fact
data euros;
Total_
Customer
. . .
Order
. . . Retail_ . . .
set orion.order_fact(where=(Order_Date
_ID _Date between '01SEP2007'd and
Price
928 . . . 04SEP2007 . . . 86.30 . . . '30SEP2007'd)
27 . . . 05SEP2007 . . . 78.40 . . . keep=Customer_ID Order_Date
31 . . . 06SEP2007 . . . 50.30 . . . Product_ID
45 . . . 06SEP2007 . . . 78.20 . . . Total_Retail_Price);
5 . . . 09SEP2007 . . . 52.50 . . . do while (not (SDate le Order_Date le
. . . . . . EDate));
. . . . . . set orion.rates;
. .
Continue until
.
.
EOF . .
end;
orion.rates for orion.order_fact. EuroPrice=Total_Retail_Price*AvgRate;
format EuroPrice Euro10.2;
SDate EDate AvgRate run;

01SEP2007 07SEP2007 0.73117

08SEP2007 14SEP2007 0.72184
15SEP2007 21SEP2007 0.71589
22SEP2007 30SEP2007 0.70725
PDV
Total_
Customer Order_ Product_ Avg Euro D
Retail_ SDate EDate _N_
_ID Date ID Rate Price
Price
11 30SEP2007 220200100002 78.20 22SEP2007 30SEP2007 0.70725 55.31 10

154
8-88 Chapter 8 Combining Data Horizontally

The Resulting Data Set

proc print data=euros;
title 'Euros';
run;

PROC PRINT Output

Euros

Order_ Total_Retail_
Obs Customer_ID Date Product_ID Price SDate EDate AvgRate EuroPrice

1 928 04SEP2007 230100600030 $86.30 01SEP2007 07SEP2007 0.73117 €63.10

2 27 05SEP2007 240500200082 $78.40 01SEP2007 07SEP2007 0.73117 €57.32
3 31 06SEP2007 220200100137 $50.30 01SEP2007 07SEP2007 0.73117 €36.78
4 45 06SEP2007 230100600015 $78.20 01SEP2007 07SEP2007 0.73117 €57.18
5 5 09SEP2007 210200500016 $52.50 08SEP2007 14SEP2007 0.72184 €37.90
6 12 18SEP2007 240200100053 $87.20 15SEP2007 21SEP2007 0.71589 €62.43
7 69 20SEP2007 210200700016 $23.50 15SEP2007 21SEP2007 0.71589 €16.82
8 24 25SEP2007 240600100102 $46.10 22SEP2007 30SEP2007 0.70725 €32.60
9 41 26SEP2007 210200600067 $134.00 22SEP2007 30SEP2007 0.70725 €94.77
10 11 30SEP2007 220200100002 $78.20 22SEP2007 30SEP2007 0.70725 €55.31

155

8.10 Multiple Choice Poll

Does SAS encounter the end-of-file marker (EOF) for
orion.rates?
a. SAS encounters the EOF, but it does not stop the
DATA step because there are two SET statements.
b. SAS encounters the EOF for orion.rates, so there are
only four observations in the resulting data set euros.
c. The DO WHILE statement prevents the data set
orion.rates from being read a fifth time, so the EOF is
never encountered.

157
 8.4 Combining Data Conditionally (Self-Study) 8-89

Using the SQL Procedure

You can use the SQL procedure to produce the same
results.
proc sql;
create table euros as
select Customer_ID, Order_Date, Product_ID,
Total_Retail_Price, SDate, EDate, AvgRate,
Total_Retail_Price*AvgRate as EuroPrice
format=Euro10.2
from orion.order_fact, orion.rates
where Order_Date between SDate and EDate;
title 'euros';
select * from euros;
quit;

 Neither data set needs to be sorted.

p308d13
159

The Resulting Data Set

PROC SQL Output
euros

Date
Order was Total Retail
placed by Price for
Customer ID Customer Product ID This Product SDate EDate AvgRate EuroPrice
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
928 04SEP2007 230100600030 $86.30 01SEP2007 07SEP2007 0.731167 €63.10
27 05SEP2007 240500200082 $78.40 01SEP2007 07SEP2007 0.731167 €57.32
31 06SEP2007 220200100137 $50.30 01SEP2007 07SEP2007 0.731167 €36.78
45 06SEP2007 230100600015 $78.20 01SEP2007 07SEP2007 0.731167 €57.18
5 09SEP2007 210200500016 $52.50 08SEP2007 14SEP2007 0.72184 €37.90
12 18SEP2007 240200100053 $87.20 15SEP2007 21SEP2007 0.715886 €62.43
69 20SEP2007 210200700016 $23.50 15SEP2007 21SEP2007 0.715886 €16.82
24 25SEP2007 240600100102 $46.10 22SEP2007 30SEP2007 0.70725 €32.60
41 26SEP2007 210200600067 $134.00 22SEP2007 30SEP2007 0.70725 €94.77
11 28SEP2007 220200100002 $78.20 22SEP2007 30SEP2007 0.72184 €55.31

160
8-90 Chapter 8 Combining Data Horizontally

Using a Hash Object

p308d14
data euros;
length SDate EDate AvgRate 8;
drop rc;
format SDate EDate Order_Date date9. EuroPrice Euro10.2;
if _N_=1 then do;
declare hash H(dataset: 'orion.rates', ordered: 'ascending');
H.definekey('SDate');
H.definedata('SDate','EDate', 'AvgRate');
H.definedone();
call missing(SDate, EDate, AvgRate);
declare hiter E('H');
end;
set orion.order_fact(where=(Order_Date between '01SEP2007'd and
'30SEP2007'd)
keep=Customer_ID Order_Date Product_ID
Total_Retail_Price);
E.first();
do until (rc ne 0);
if SDate <= Order_Date <= EDate then do;
EuroPrice=Total_Retail_Price * AvgRate;
output;
leave;
end;
else if SDate > Order_Date then leave;
rc=E.next();
end;
run;
 8.4 Combining Data Conditionally (Self-Study) 8-91

Which Technique Should You Use?

Two SET Statements SQL Procedure Hash and Hiter Objects
availability of all DATA easy to code availability of all DATA
step functionality step functionality

fast, sequential no sorting or in-memory lookup table

processing of both indexing needed
data sets

 Benchmark your data to determine which is most

efficient.

162
8-92 Chapter 8 Combining Data Horizontally

Exercises

Level 1

10. Combining Two Data Sets Conditionally Using the SQL Procedure
The data set orion.ages_mod contains information about age groups.
Listing of orion.ages_mod
orion.ages_mod SAS Data Set

First_
Description Age Last_Age

15-29 years 15 30
30-44 years 30 45
45-59 years 45 60
60-75 years 60 75

The data set orion.customer contains information about customers.

Partial Listing of orion.customer
orion.customer SAS Data Set
(Partial Output)

Personal_ Customer_ Customer_

Customer_ID Country Gender ID Customer_Name FirstName LastName

4 US M James Kvarniq James Kvarniq

5 US F Sandrina Stephano Sandrina Stephano
9 DE F Cornelia Krahl Cornelia Krahl
10 US F Karen Ballinger Karen Ballinger
11 DE F Elke Wallstab Elke Wallstab

Birth_ Street_ Customer_

Date Customer_Address Street_ID Number Type_ID

27JUN1974 4382 Gralyn Rd 9260106519 4382 1020

09JUL1979 6468 Cog Hill Ct 9260114570 6468 2020
27FEB1974 Kallstadterstr. 9 3940106659 9 2020
18OCT1984 425 Bryant Estates Dr 9260129395 425 1040
16AUG1974 Carl-Zeiss-Str. 15 3940108592 15 1040

a. Use the SQL procedure to create a data set named age_groups that contains the customer ID,
name, age, and age group (the variable Description in the orion.ages_mod SAS data set) as of
January 1, 2008. Order the data by Customer_ID.
Hint: The following calculates the customer age:

int(yrdif(Birth_Date,'01Jan2008'd, 'ACT/ACT'))
 8.4 Combining Data Conditionally (Self-Study) 8-93

b. Print the first five observations of the age_groups data set.

PROC PRINT Output
age_groups
(Partial Output)

Obs Customer_ID Customer_Name Age Description

1 4 James Kvarniq 33 30-44 years

2 5 Sandrina Stephano 28 15-29 years
3 9 Cornelia Krahl 33 30-44 years
4 10 Karen Ballinger 23 15-29 years
5 11 Elke Wallstab 33 30-44 years

Level 2

11. Combining Two Data Sets Conditionally Using the DATA Step DO Loop
The data set orion.ages_mod contains information about age groups.
Listing of orion.ages_mod
orion.ages_mod SAS Data Set

First_
Description Age Last_Age

15-29 years 15 30
30-44 years 30 45
45-59 years 45 60
60-75 years 60 75

The data set orion.customer contains information about customers.

Partial Listing of orion.customer
orion.customer SAS Data Set
(Partial Output)

Personal_ Customer_ Customer_

Customer_ID Country Gender ID Customer_Name FirstName LastName

4 US M James Kvarniq James Kvarniq

5 US F Sandrina Stephano Sandrina Stephano
9 DE F Cornelia Krahl Cornelia Krahl
10 US F Karen Ballinger Karen Ballinger
11 DE F Elke Wallstab Elke Wallstab

Birth_ Street_ Customer_

Date Customer_Address Street_ID Number Type_ID

27JUN1974 4382 Gralyn Rd 9260106519 4382 1020

09JUL1979 6468 Cog Hill Ct 9260114570 6468 2020
27FEB1974 Kallstadterstr. 9 3940106659 9 2020
18OCT1984 425 Bryant Estates Dr 9260129395 425 1040
16AUG1974 Carl-Zeiss-Str. 15 3940108592 15 1040
8-94 Chapter 8 Combining Data Horizontally

a. Use the DATA step and a DO loop to create a data set named age_groups that contains the
customer ID, name, age, and age group (the variable Description in the orion.ages_mod
SAS data set) as of January 1, 2008.
b. Print the first five observations of the age_groups data set.
PROC PRINT Output
age_groups
(Partial Output)

Obs Customer_ID Customer_Name Age

1 2806 Raedene Van Den Berg 15

2 13 Markus Sepke 15
3 2550 Sanelisiwe Collier 15
4 46966 Lauren Krasowski 17
5 11171 Bill Cuddy 17

Level 3

12. Combining Two Data Sets Conditionally Using the DATA Step Hash Object
The data set orion.ages_mod contains information about age groups.
Listing of orion.ages_mod
orion.ages_mod SAS Data Set

First_
Description Age Last_Age

15-29 years 15 30
30-44 years 30 45
45-59 years 45 60
60-75 years 60 75

The data set orion.customer contains information about customers.

Partial Listing of orion.customer
orion.customer SAS Data Set
(Partial Output)

Personal_ Customer_ Customer_

Customer_ID Country Gender ID Customer_Name FirstName LastName

4 US M James Kvarniq James Kvarniq

5 US F Sandrina Stephano Sandrina Stephano
9 DE F Cornelia Krahl Cornelia Krahl
10 US F Karen Ballinger Karen Ballinger

Birth_ Street_ Customer_

Date Customer_Address Street_ID Number Type_ID

27JUN1974 4382 Gralyn Rd 9260106519 4382 1020

09JUL1979 6468 Cog Hill Ct 9260114570 6468 2020
27FEB1974 Kallstadterstr. 9 3940106659 9 2020
18OCT1984 425 Bryant Estates Dr 9260129395 425 1040
 8.4 Combining Data Conditionally (Self-Study) 8-95

a. Use the DATA step and a hash object to create a data set named age_groups that contains the
customer ID, name, age, and age group (the variable Description in the orion.ages_mod
SAS data set) as of January 1, 2008.
b. Print the first five observations of the age_groups data set.
PROC PRINT Output
age_groups
(Partial Output)

Obs Description Customer_ID Customer_Name Age

1 30-44 years 4 James Kvarniq 33

2 15-29 years 5 Sandrina Stephano 28
3 30-44 years 9 Cornelia Krahl 33
4 15-29 years 10 Karen Ballinger 23
5 30-44 years 11 Elke Wallstab 33
8-96 Chapter 8 Combining Data Horizontally

8.5 Chapter Review

Chapter Review
1. Given the following input data, what is one difference
in data sets created by the default DATA step MERGE
and the default SQL procedure inner join?
one two
X Y X Z
1 a 1 f
2 b 3 t
3 c 4 w

165 continued...

Chapter Review
2. When data sets are combined using the SET/SET
KEY= syntax, how is the data set named in the first
SET statement read?

3. When data sets are combined using the SET/SET

KEY= syntax, how is the data set named in the
second SET statement read?

167 continued...
 8.5 Chapter Review 8-97

Chapter Review
4. If the following program is used to combine the
summary data set containing one observation with the
detail data set containing fifty observations, how many
observations are in the data set combined?
data combined;
set summary;
set detail;
run;

169
8-98 Chapter 8 Combining Data Horizontally

8.6 Solutions

Solutions to Exercises
1. Merging or Joining Three Data Sets
a. Combine the three data sets to create a data set named purchases that contains the customer
name, product name, and supplier name for the customers in the orion.order_fact data set.
p308s01
/* Merge Solution */

proc sort data=orion.order_fact(keep=Customer_ID Product_ID)

out=order_fact;
by Customer_ID;
run;

data temp;
merge order_fact(in=O)
orion.customer_dim(keep=Customer_ID Customer_Name
in=C);
by Customer_ID;
if O and C;
run;

proc sort data=temp;

by Product_ID;
run;

data purchases;
keep Customer_Name Product_Name Supplier_Name;
merge temp(in=T)
orion.product_dim(keep=Product_ID Product_Name
Supplier_Name
In=P);
by Product_ID;
if P and T;
run;

proc print data=purchases(obs=5);

title 'Partial purchases Data Set';
run;
(Continued on the next page.)
 8.6 Solutions 8-99

/* PROC SQL Solution */

proc sql;
create table purchases as
select Customer_Name,
Product_Name,
Supplier_Name
from orion.order_fact,
orion.product_dim,
orion.customer_dim
where order_fact.Customer_ID=customer_dim.Customer_ID
and order_fact.Product_ID=product_dim.Product_ID
order by order_fact.Product_ID;
quit;
b. Order the data by Product_ID and print the first five observations of the purchases data set.
p308s01
proc print data=purchases(obs=5);
title 'Partial Purchases Data Set';
run;
2. Merging or Joining Data to Create Multiple Data Sets
Combine the three data sets to create the following data sets:
• a data set named no_purchases that contains the customers who did not make any purchases
• a data set named purchases that contains the customer name, product name, and supplier name for
those customers in the orion.order_fact data set
• a data set named no_products that contains the product names and suppliers for products that were
not purchased
p308s02
/* Merge Solution */
proc sort data=orion.order_fact out=order_fact;
by Customer_ID;
run;

data temp no_purchases(keep=Customer_ID Customer_Name);

merge order_fact(in=O)
orion.customer_dim(keep=Customer_IDCustomer_Name in=C);
by Customer_ID;
if O and C then output temp;
else if C and not O then output no_purchases;
run;

proc sort data=temp;

by Product_ID;
run;
(Continued on the next page.)
8-100 Chapter 8 Combining Data Horizontally

data purchases
no_products(keep=Product_ID Product_Name
Supplier_Name);
merge temp(in=T)
orion.product_dim(keep=Product_ID Product_Name
Supplier_Name
in=P);
by Product_ID;
if P and T then output purchases;
else if P and not T then output no_products;
run;

proc print data=no_purchases;

title 'no_purchases Data Set';
run;

proc print data=purchases(obs=5);

title 'Partial purchases Data Set';
run;

proc print data=no_products(obs=5);

title 'Partial no_products Data Set';
run;

/* PROC SQL Solution */

proc sql;
create table no_purchases as
select Customer_ID,
Customer_Name
from orion.customer_dim
where customer_dim.Customer_ID not in
(select Customer_ID from orion.order_fact);
create table no_products as
select Product_ID, Product_Name
from orion.product_dim
where product_dim.Product_ID not in
(select Product_ID from orion.order_fact);
create table purchases as
select order_fact.*,
Customer_Name,
Product_Name,
Supplier_Name
from orion.order_fact,
orion.product_dim,
orion.customer_dim
where order_fact.Customer_ID=customer_dim.Customer_ID
and order_fact.Product_ID=product_dim.Product_ID
order by order_fact.Product_ID;
quit;
(Continued on the next page.)
 8.6 Solutions 8-101

proc print data=no_products(obs=5);

title 'Partial no_products Data Set';
run;

proc print data=purchases(obs=5);

title 'Partial purchases Data Set';
run;

proc print data=no_purchases;

title 'no_purchases Data Set';
run;
3. Merging or Joining Multiple Data Sets
Create a data set named manager_names that contains the Employee_ID variable, the six
Manager_ID variables, and the six manager names.
p308s03
* PROC SQL Solution */
proc sql;
create table Manager_Names as
select e.Employee_ID,
Manager_Level1,
Manager_Level2,
Manager_Level3,
Manager_Level4,
Manager_Level5,
Manager_Level6,
m1.Employee_Name as Manager1_Name,
m2.Employee_Name as Manager2_Name,
m3.Employee_Name as Manager3_Name,
m4.Employee_Name as Manager4_Name,
m5.Employee_Name as Manager5_Name,
m6.Employee_Name as Manager6_Name
from orion.organization_dim as e
left join orion.employee_addresses as m1
on e.Manager_Level1=m1.Employee_ID
left join orion.employee_addresses as m2
on e.Manager_Level2=m2.Employee_ID
left join orion.employee_addresses as m3
on e.Manager_Level3=m3.Employee_ID
left join orion.employee_addresses as m4
on e.Manager_Level4=m4.Employee_ID
left join orion.employee_addresses as m5
on e.Manager_Level5=m5.Employee_ID
left join orion.employee_addresses as m6
on e.Manager_Level6=m6.Employee_ID
order by e.Employee_ID;
quit;
(Continued on the next page.)
8-102 Chapter 8 Combining Data Horizontally

proc print data=manager_names(firstobs=420);

title 'Partial Manager_Names Data';
title2 'FirstObs=420';
run;

/* Merge Solution */

proc sort data=orion.employee_addresses

out=emp_addresses(rename=(Employee_ID=Manager_Level1
Employee_Name=Manager1_Name));
by Employee_ID;
run;

proc sort data=orion.organization_dim

out=man1(keep=Employee_ID Manager_Level1 - Manager_Level6);
by Manager_Level1;
run;

data manager1;
merge man1(in=M)
emp_addresses(keep=Manager_Level1 Manager1_Name);
by Manager_Level1;
if M;
run;

proc sort data=manager1 out=man2;

by Manager_Level2;
run;

data manager2;
merge man2(in=M)
emp_addresses(rename=(Manager_Level1=Manager_Level2
Manager1_Name=Manger2_Name)
keep=Manager_Level1 Manager1_Name);
by Manager_Level2;
if M;
run;

proc sort data=manager2 out=man3;

by Manager_Level3;
run;
(Continued on the next page.)
 8.6 Solutions 8-103

data manager3;
merge man3(in=M)
emp_addresses(rename=(Manager_Level1=Manager_Level3
Manager1_Name=Manger3_Name)
keep=Manager_Level1 Manager1_Name);
by Manager_Level3;
if M;
run;

proc sort data=manager3 out=man4;

by Manager_Level4;
run;

data manager4;
merge man4(in=M)
emp_addresses(rename=(Manager_Level1=Manager_Level4
Manager1_Name=Manger4_Name)
keep=Manager_Level1 Manager1_Name);
by Manager_Level4;
if M;
run;

proc sort data=manager4 out=man5;

by Manager_Level5;
run;

data manager5;
merge man5(in=M)
emp_addresses(rename=(Manager_Level1=Manager_Level5
Manager1_Name=Manger5_Name)
keep=Manager_Level1 Manager1_Name);
by Manager_Level5;
if M;
run;

proc sort data=manager5 out=man6;

by Manager_Level6;
run;

data manager_names;
merge man6(in=M)
emp_addresses(rename=(Manager_Level1=Manager_Level6
Manager1_Name=Manger6_Name)
keep=Manager_Level1 Manager1_Name);
by Manager_Level6;
if M;
run;
(Continued on the next page.)
8-104 Chapter 8 Combining Data Horizontally

proc sort data=manager_names;

by Employee_ID;
run;

proc print data=manager_names(firstobs=420);

title 'Partial Manager_Names Data';
title2 'FirstObs=420';
run;
4. Combining Data Sets Using an Index to Create One Data Set
a. Create a SAS data set named sales_emps by using an index on Employee_ID to combine the two
data sets, orion.salesstaff and orion.organization_dim. Check the SAS log to ensure that you do
not have any data errors. Read only the variables Employee_ID, Department, Section, and
Org_Group from orion.organization_dim.
p308s04
proc datasets lib=orion nolist;
modify organization_dim;
index create Employee_ID;
quit;

data sales_emps;
set orion.salesstaff;
set orion.organization_dim(keep=Employee_ID Department
Section Org_Group)
key=Employee_ID;
if _IORC_=0;
run;
b. Print the first five observations of the sales_emps SAS data set.
p308s04
proc print data=sales_emps(obs=5);
title 'Sales Employee Data';
title2 '(Partial Output)';
run;
 8.6 Solutions 8-105

5. Combining Data Sets Using an Index to Monitor Data Integrity

a. Create a simple index on the variable Product_ID in the data set orion.shoe_prices.
b. Read only the variables Product_ID, Product_Name, Supplier_Name, and
Mfg_Suggested_Retail_Price from orion.shoe_vendors.
Hint: There is a permanent format assigned to the Supplier_Country variable. To avoid a syntax
error, use the NOFMTERR system option.
c. Read only the variables Product_ID, Total_Retail_Price, CostPrice_Per_Unit from
orion.shoe_prices.
The shoes data set should have the price information for the shoe products.
d. The errors data set should contain data that is in orion.shoe_vendors, which is not in the
orion.shoe_prices data. The errors data set should contain only the variables Product_ID,
Product_Name, and Supplier_Name.

 The errors data can then be used to determine why these vendors do not have
observations in price_list.
e. Delete the Product_ID index on the data set orion.shoe_prices.
p308s05
proc datasets lib=orion nolist;
modify shoe_prices;
index create Product_ID;
run;
quit;

/**************************************/
/* If you keep the supplier_country */
/* variable, uncomment and submit */
/* the following options statement */
/* to avoid an error */
/**************************************/

*options nofmterr;

data shoes errors(keep=Product_ID Product_Name Supplier_Name);

set orion.shoe_vendors(keep=Product_ID Product_Name
Supplier_Name Mfg_Suggested_Retail_Price);
set orion.shoe_prices(keep=Product_ID Total_Retail_Price
CostPrice_per_Unit)
key=Product_ID;
if _IORC_=0 then output shoes;
else do;
_ERROR_=0;
output errors;
end;
run;
(Continued on the next page.)
8-106 Chapter 8 Combining Data Horizontally

proc datasets lib=orion nolist;

modify shoe_prices;
index delete Product_ID;
run;
quit;
f. Print the first five observations of the shoes SAS data set.
p308s05
proc print data=shoes(obs=5);
title 'Shoe Data';
title2 '(Partial Output)';
run;
g. Print the observations of the errors SAS data set.
p308s05
proc print data=errors;
title 'The errors Data';
run;
6. Combining Data Sets Using an Index and Using the Macro Facility to Monitor Errors
a. Create a data set named processed_orders that contains the variables from
orion.first_internet_order and a variable named Comment. Use the index on the variable
Order_ID to retrieve the matching observation from orion.internet.
b. The variable Comment has the value Order has been processed if the Order_ID is
in both orion.first_internet_order and orion.internet. The value is Order has not been
processed if the Customer_ID is not in both data sets.

c. Use the %SYSRC AUTOCALL macro described in the reference information in this chapter. In
addition, refer to SAS OnlineDoc by following the path shown below:
Support & Training Ö Knowledge Base Ö Documentation Ö Base SAS Ö
SAS 9.2 Macro Language: Reference Ö Macro Language Dictionary Ö AutoCall Macros
 8.6 Solutions 8-107

p308s06
data processed_orders;
set orion.first_internet_order;
set orion.internet key=Order_ID;
length Comment $30;
select (_IORC_);
when (%sysrc(_sok)) do;
Comment='Order has been processed.';
output;
end;
when (%sysrc(_dsenom)) do;
_ERROR_=0;
Comment='Order has not been processed.';
output;
end;
otherwise;
end;
run;
d. Print the first 10 observations of processed_orders.
p308s06
proc print data=processed_orders(obs=10);
title 'Internet Orders';
title2 '(Partial Output)';
run;
7. Combining Summary Data Containing an Average with Detail Data
a. Calculate the average age of all customers.
b. Create a SAS data set named age_dif, which combines the average age of all customers with the
orion.customer_dim data set in order to determine the difference between each customer’s age
and the average for all customers. (You can use any method presented in this section.)
p308s07
/* Using PROC SUMMARY and the DATA step */

proc summary data=orion.customer_dim;

var Customer_Age;
output out=average mean=AvgAge;
run;

data age_dif;
if _N_=1 then set average(keep=AvgAge);
set orion.customer_dim(keep=Customer_ID Customer_Age);
Age_Difference=Customer_Age - AvgAge;
run;
(Continued on the next page.)
8-108 Chapter 8 Combining Data Horizontally

/* Using PROC SUMMARY and PROC SQL */

proc summary data=orion.customer_dim;

var Customer_Age;
output out=average mean=AvgAge;
run;

proc sql;
create table age_dif as
select AvgAge,
Customer_ID,
Customer_Age,
Customer_Age - AvgAge as Age_Difference
from orion.customer_dim,
average;
quit;

/* Using PROC SQL only */

proc sql;
create table age_dif as
select mean(Customer_Age) as AvgAge,
Customer_ID,
Customer_Age,
Customer_Age - calculated AvgAge as Age_Difference
from orion.customer_dim;
quit;

/* Using the DATA step only */

data age_dif;
drop i Tot_Age;
if _N_=1 then do i=1 to TotObs;
set orion.customer_dim(keep=Customer_Age) nobs=TotObs;
Tot_Age + Customer_Age;
end;
set orion.customer_dim(keep=Customer_ID Customer_Age);
AvgAge=Tot_Age / TotObs;
Age_Difference=Customer_Age - AvgAge;
run;
c. Print the first five observation of the age_dif SAS data set.
p308s07
proc print data=age_dif(obs=5);
var AvgAge Customer_ID Customer_Age Age_Difference;
title 'The age_dif Data Set';
title2 '(Partial Output)';
run;
 8.6 Solutions 8-109

8. Combining Summary Data Containing a Total with Detail Data

a. Select any method to create a SAS data set named compare.
p308s08
/* Using PROC SUMMARY and the DATA step */

data donations;
set orion.employee_donations;
Total_Donation=sum(of Qtr1 - Qtr4);
run;

proc summary data=donations;

var Total_Donation;
output out=totals mean=Avg_Donation;
run;

data compare;
if _N_=1 then set totals;
set donations;
Difference=Total_Donation - Avg_Donation;
run;

/* Using PROC SUMMARY and PROC SQL*/

proc sql;
create table donations as
select Employee_ID,
Qtr1,
Qtr2,
Qtr3,
Qtr4,
Recipients,
Paid_By,
sum(Qtr1, Qtr2, Qtr3, Qtr4) as Total_Donation
from orion.employee_donations;

proc summary data=donations;

var Total_Donation;
output out=totals mean=Avg_Donation;
run;
(Continued on the next page.)
8-110 Chapter 8 Combining Data Horizontally

proc sql;
create table compare as
select Avg_Donation,
donations.*,
Total_Donation - Avg_Donation as Difference
from totals,
donations;
quit;
/* Using PROC SQL only */
proc sql;
create table compare as
select mean(sum(Qtr1, Qtr2, Qtr3, Qtr4)) as Avg_Donation,
employee_donations.*,
sum(Qtr1, Qtr2, Qtr3, Qtr4) as Total_Donation,
calculated Total_Donation - calculated Avg_Donation
as Difference
from orion.employee_donations;
quit;

/* Using the DATA step only */

data compare;
drop i;
if _N_=1 then do i=1 to TotObs;
set orion.employee_donations(keep=Qtr1 - Qtr4)
nobs=TotObs;
Total + sum(of Qtr1 - Qtr4);
end;
set orion.employee_donations;
Total_Donation=sum(of Qtr1 - Qtr4);
Avg_Donation=Total / TotObs;
Difference=Total_Donation-Avg_Donation;
run;
b. Print the first five observations of the compare SAS data set.
p308s08
proc print data=compare(obs=5);
var Avg_Donation Employee_ID Qtr1 Qtr2 Qtr3 Qtr4
Recipients Paid_By Total_Donation Difference;
title 'The compare Data Set';
title2 '(Partial Output)';
run;
 8.6 Solutions 8-111

9. Combining Summary Data Containing a Weighted Average and Detail Data

a. Select any method to create a SAS data set named products by performing the following tasks:
p308s09
/* Using PROC SUMMARY and the DATA STEP */

proc summary data=orion.order_fact;

var CostPrice_Per_Unit;
weight Quantity;
output out=totals sum=Total_Cost;
run;

proc sort data=orion.order_fact out=order_fact;

by Product_ID;
run;

data products(keep=Customer_ID CostPrice_Per_Unit Quantity

Percent Product_Name);
if _N_=1 then set totals(keep=Total_Cost);
merge order_fact(keep=Customer_ID Product_ID
CostPrice_Per_Unit Quantity in=O)
orion.product_dim(keep=Product_ID Product_Name in=P);
by Product_ID;
if O and P;
Percent=(CostPrice_Per_Unit * Quantity) / Total_Cost;
format Percent percent9.3;
run;

/* Using PROC SUMMARY and PROC SQL */

proc summary data=orion.order_fact;

var CostPrice_Per_Unit;
weight Quantity;
output out=totals sum=Total_Cost;
run;

proc sql;
create table products as
select Customer_ID,
CostPrice_Per_Unit,
Quantity,
Product_Name,
(Quantity * CostPrice_Per_Unit) / Total_Cost as
Percent format=percent9.3
from totals,
orion.order_fact,
orion.product_dim
where order_fact.Product_ID=product_dim.Product_ID;
quit;
(Continued on the next page.)
8-112 Chapter 8 Combining Data Horizontally

/* Using PROC SQL only */

proc sql;
create table products as
select Customer_ID,
CostPrice_Per_Unit,
Quantity,
Product_Name,
(Quantity * CostPrice_Per_Unit)/
sum(Quantity * CostPrice_Per_Unit)as Percent
format=percent9.3
from orion.order_fact,
orion.product_dim
where order_fact.Product_ID=product_dim.Product_ID;
quit;

/* Using the DATA step only */

proc sort data=orion.order_fact out=order_fact;

by Product_ID;
run;

data products(keep=Customer_ID CostPrice_Per_Unit Quantity

Percent Product_Name);
if _N_=1 then do i=1 to TotObs;
set orion.order_fact nobs=TotObs;
Total_Cost + (Quantity * CostPrice_Per_Unit);
end;
merge order_fact(keep=Customer_ID Product_ID
CostPrice_Per_Unit Quantity
in=O)
orion.product_dim(keep=Product_ID Product_Name
in=P);
by Product_ID;
if O and P;
Percent=(CostPrice_Per_Unit * Quantity) / Total_Cost;
format Percent percent9.3;
run;
b. Print the first five observations of the products SAS data set.
p308s09
proc print data=products(obs=5);
var Customer_ID Quantity CostPrice_Per_Unit Product_Name
Percent;
title 'The products Data Set';
title2 '(Partial Output)';
run;
 8.6 Solutions 8-113

10. Combining Two Data Sets Conditionally Using the SQL Procedure
a. Use the SQL procedure to create a data set named age_groups.
p308s10
proc sql;
create table age_groups as
select Customer_ID,
Customer_Name,
int(yrdif(Birth_Date, '01Jan2008'd, 'ACT/ACT')) as Age,
Description
from orion.customer,
orion.ages_mod
where calculated Age between First_Age and Last_Age
order by Customer_ID;
quit;
b. Print the first five observations of the age_groups data set.
p308s10
proc print data=age_groups(obs=5);
title 'age_groups';
title2 '(Partial Output)';
run;
11. Combining Two Data Sets Conditionally Using the DATA Step DO Loop
a. Use the DATA step and a DO loop to create a data set named age_groups.
p308s11
proc sort data=orion.customer(keep=Customer_ID Birth_Date
Customer_Name)
out=customer;
by descending Birth_Date;
run;

data age_groups;
keep Customer_ID Customer_Name Age Description;
set customer;
Age=int(yrdif(Birth_Date, '01Jan2008'd, 'ACT/ACT'));
do while (not (First_Age le Age lt Last_Age));
set orion.ages_mod;
end;
run;
b. Print the first five observations of the age_groups data set.
p308s11
proc print data=age_groups(obs=5);
title 'age_groups';
title2 '(Partial Output)';
run;
8-114 Chapter 8 Combining Data Horizontally

12. Combining Two Data Sets Conditionally Using the DATA Step Hash Object
a. Use the DATA step and a hash object to create a data set named age_groups that contains the
customer ID, name, age, and age group (the variable Description in the orion.ages_mod
SAS data set) as of January 1, 2008.
p308s12
/* Using the DATA Step Hash Object */

data age_groups;
keep Customer_ID Customer_Name Age Description;
if _N_=1 then do;
if 0 then set orion.ages_mod;
declare hash AG(dataset: 'orion.ages_mod',
ordered: 'ascending');
AG.definekey('First_Age');
AG.definedata('First_Age', 'Last_Age', 'Description');
AG.definedone();
declare hiter A('AG');
end;

set orion.customer(keep=Customer_ID Birth_Date

Customer_Name);
Age=int(yrdif(Birth_Date, '01Jan2008'd, 'ACT/ACT'));
A.first();
do until (rc ne 0);
if First_Age <= Age < Last_Age then do;
output;
leave;
end;
else if First_Age > Age then leave;
rc=A.next();
end;
run;

proc print data=age_groups(obs=5);

title 'age_groups';
title2 '(Partial Output)';
run;
(Continued on the next page.)
 8.6 Solutions 8-115

/* alternative solution */

data age_groups(keep=Description Customer_ID Customer_Name

Age);
if _N_=1 then do;
if 0 then set orion.ages_mod;
declare hash AG (dataset:'orion.ages_mod',
ordered: 'ascending');
ag.definekey('First_Age');
ag.definedata('Description');
ag.definedata('First_Age', 'Last_Age');
ag.definedone();
declare hiter HAG('AG');
end;
set orion.customer(keep=Customer_ID Customer_Name
Birth_Date);
Age=int(yrdif(Birth_Date, '01JAN2008'd, 'ACT/ACT'));
rc=HAG.first();
do until (First_Age le Age le Last_Age);
rc=HAG.next();
end;
run;
b. Print the first five observations of the age_groups data set.
p308s12
proc print data=age_groups(obs=5);
title 'age_groups';
title2 '(Partial Output)';
run;
8-116 Chapter 8 Combining Data Horizontally

Solutions to Student Activities (Polls/Quizzes)

8.01 Multiple Choice Poll – Correct Answer

By default, how does the DATA step perform a merge?
a. sequentially
b. creates a Cartesian product

8.02 Multiple Choice Poll – Correct Answer

By default, how does the SQL procedure perform an inner
join?
a. sequentially
b. creates a Cartesian product

19
 8.6 Solutions 8-117

8.03 Multiple Choice Poll – Correct Answer

If you use a DATA step MERGE to combine the data sets,
how many observations are read from the larger data set,
orion.customer_dim_more?
a. 38
b. 1,500

The MERGE statement reads all observations from all

of the input data sets.

If there is a great disparity in size between the input

data sets, merging might not be the best technique
for combining them.

38

8.04 Quiz – Correct Answer

Why do you not want this observation output to
catalog_customers?
Partial PDV
Total_
Customer_ Order_ Customer_
Quantity Retail_ ... ... D _IORC_ D _N_
ID ID Name
Price
Sandrina
15 1240080101 3 216.50 ...
Stephano
... 1230015 2

Data from the previous observation is retained in the

PDV. If this observation is output, the variables
Customer_Country through Customer_Age would be
incorrect for Customer_ID=15.

54
8-118 Chapter 8 Combining Data Horizontally

8.05 Quiz – Correct Answer

Open and submit the program p308a01.
1. What messages do you see in your SAS log?
The first message is as follows:
Customer_ID=15 Order_ID=1240080101 Quantity=3
Total_Retail_Price=$216.50 Customer_Country=US
Customer_Gender=F Customer_Name=Sandrina Stephano
Customer_FirstName=Sandrina
Customer_LastName=Stephano Customer_BirthDate=09JUL1979
Customer_Age_Group=15-30 years
Customer_Type=Orion Club Gold members medium activity
Customer_Group=Orion Club Gold members
Customer_Age=28 _ERROR_=1 _IORC_=1230015 _N_=2

By default, there is a maximum of 20 error messages

printed in the log.

60

8.05 Quiz – Correct Answer

Open and submit the program p308a01.
1. What messages do you see in your SAS log?
2. What is the value of _ERROR_?
_ERROR_=1

When _ERROR_=1, the contents of the PDV are

printed in the log.

61
 8.6 Solutions 8-119

8.05 Quiz – Correct Answer

Open and submit the program p308a01.
1. What messages do you see in your SAS log?
2. What is the value of _ERROR_?
3. Replace the ELSE statement with the following ELSE
DO group:
else do;
_ERROR_=0;
output errors;
end;
4. Resubmit the program and look at the log.
5. Why are there no messages now?
You forced _ERROR_ to be 0 with the assignment
statement _ERROR_=0. Therefore, the data error
62
messages are not printed in the log.

8.06 Quiz – Correct Answer

Open and submit the program p308a02.
1. How many observations are in the resulting data set?
Only one observation
2. Why?
When the statement
set summary(keep=GrandTot);
executes a second time, SAS encounters the end
of file in the summary data set because that data
set only has one observation in it.

98 continued...
8-120 Chapter 8 Combining Data Horizontally

8.06 Quiz – Correct Answer

Open and submit the program p308a02.
3. How did you get 53 observations in the program
p308d06?
The statement set summary(keep=GrandTot); does
not execute a second time because execution
of the SET statement is controlled by the condition
if _N_=1.

99

8.07 Poll – Correct Answer

Can the DATA step merge be used for this task?
€ Yes
€ No

119
 8.6 Solutions 8-121

8.08 Quiz – Correct Answer

Why do you have to use a WHERE= data set option
rather than a WHERE statement to subset by
Order_Date?

The variable Order_Date is only in the data set

orion.order_fact.

128

8.09 Quiz – Correct Answer

What variables are set to missing at the top of the DATA
step?

Only the variable EuroPrice is set to missing at the

top of the DATA step because the other variable
values come from an existing SAS data set.

140
8-122 Chapter 8 Combining Data Horizontally

8.10 Multiple Choice Poll – Correct Answer

Does SAS encounter the end-of-file marker (EOF) for
orion.rates?
a. SAS encounters the EOF, but it does not stop the
DATA step because there are two SET statements.
b. SAS encounters the EOF for orion.rates, so there are
only four observations in the resulting data set euros.
c. The DO WHILE statement prevents the data set
orion.rates from being read a fifth time, so the EOF is
never encountered.

158
 8.6 Solutions 8-123

Solutions to Chapter Review

Chapter Review – Correct Answers

1. Given the following input data, what is one difference
in data sets created by the default DATA step MERGE
and the default SQL procedure inner join?
one two
X Y X Z
1 a 1 f
2 b 3 t
3 c 4 w

By default, the data created by the MERGE

statement contains the matching and nonmatching
rows. The data created by the PROC SQL inner
join contains only the matches.
166 continued...

Chapter Review – Correct Answers

2. When data sets are combined using the SET/SET
KEY= syntax, how is the data set named in the first
SET statement read?
sequentially

3. When data sets are combined using the SET/SET

KEY= syntax, how is the data set named in the
second SET statement read?
direct access

168 continued...
8-124 Chapter 8 Combining Data Horizontally

Chapter Review – Correct Answers

4. If the following program is used to combine the
summary data set containing one observation with the
detail data set containing 50 observations, how many
observations are in the data set combined?
data combined;
set summary;
set detail;
run;

170
Chapter 9 Sorting SAS Data Sets

9.1 Using the SORT Procedure ........................................................................................... 9-3

Demonstration: Using the EQUALS and NOEQUALS Options ........................................... 9-21

Demonstration: Using the NUMERIC_COLLATION= Option .............................................. 9-27

Exercises .............................................................................................................................. 9-30

9.2 BY-Group Processing (Self-Study) ............................................................................. 9-33

Exercises .............................................................................................................................. 9-53

9.3 Chapter Review............................................................................................................. 9-58

9.4 Solutions ....................................................................................................................... 9-59

Solutions to Exercises .......................................................................................................... 9-59

Solutions to Student Activities (Polls/Quizzes) ..................................................................... 9-66

Solutions to Chapter Review ................................................................................................ 9-74

9-2 Chapter 9 Sorting SAS Data Sets
 9.1 Using the SORT Procedure 9-3

9.1 Using the SORT Procedure

Objectives
„ List the reasons for sorting data.
„ Define the SAS sort.
„ Define threading.
„ Calculate the workspace and library space
required to sort a SAS data file.
„ Allocate sort workspace.
„ Use the EQUALS|NOEQUALS option.
„ Use the SORTEDBY= option.
„ Use the PRESORTED option.
„ Change the collating sequence of the SORT
procedure.

Reasons for Sorting Data

Data is sorted to accomplish the following:
„ reorder the data for reporting

Create a report with employees listed in alphabetical order by

last name.

„ store ordered data to reduce data retrieval time

A WHERE statement executes faster if the data is sorted
by the variables used in the WHERE expression.

„ enable BY-group processing in both DATA and

PROC steps
Create individual reports for each employee.

6
9-4 Chapter 9 Sorting SAS Data Sets

Using the SAS Sorting Utility

The SORT procedure has the following characteristics:
„ supplied by SAS for all operating environments

„ executes in memory up to the limit imposed by the

SORTSIZE= option
„ minimizes the use of external storage

„ executes in parallel using multiple threads

Threading Terminology
In SAS®9, the SORT procedure is multi-threaded.
thread a single, independent flow of control
through a program or within a
process
symmetric computers with multiple CPUs that
multiprocessing share the same memory and a
machines thread-enabled operating system,
(SMPs) providing the ability to spawn and
process multiple threads
simultaneously
parallel multiple units of work scheduled for
processing concurrent execution by the
operating system
8
 9.1 Using the SORT Procedure 9-5

Parallel Processing with Four Threads

Thread 1 Thread 2 Thread 3 Thread 4
Data file partitioned
into chunks 1 2 3 4

SORT

Partial results

Collate process

Multi-Threaded Processing
Threading can be enabled or disabled for the following
Base SAS procedures:
„ PROC MEANS/SUMMARY

„ PROC REPORT

„ PROC SORT

„ PROC SQL (GROUP BY and ORDER BY)

„ PROC TABULATE

10

 When you benchmark using the threaded procedures, use the real-time statistic rather than the
CPU-time statistic. The back-end collating process to re-create the single data set might result in
an increase in total CPU time, while reducing wall-clock time (time from submission of code for
execution to return of results).
9-6 Chapter 9 Sorting SAS Data Sets

Threaded Procedures in Base SAS

Threaded processing can be controlled using either of the
following:
„ SAS system option THREADS|NOTHREADS

OPTIONS THREADS | NOTHREADS;

„ THREADS|NOTHREADS option in the PROC

statement

PROC SORT DATA=SAS-data-set THREADS | NOTHREADS;

11

 If the TAGSORT option is used with PROC SORT, threading is disabled. The TAGSORT option
stores only the BY variables and the observation numbers (named tags) in temporary files. At the
completion of the sorting process, PROC SORT uses the tags to retrieve records from the input
data set in sorted order.

9.01 Quiz
Open and submit the program p309a01.
How many CPUs are available in your SAS session?

proc options option=cpucount;

run;

13
 9.1 Using the SORT Procedure 9-7

Threaded Procedures in Base SAS

The number of CPUs available for SAS to use can be
controlled with the CPUCOUNT= SAS system option.
General form of the CPUCOUNT= SAS system option:

OPTIONS CPUCOUNT=ACTUAL | 1-1024;

ACTUAL the number of physical processors available

(Default) when the option is set
1-1024 the number of CPUs that SAS will assume
are available for use by threaded-enabled
applications

15

 The SAS Administrator might limit the number of CPUs that are available for SAS processing, so
the value ACTUAL might be less than the total number of CPUs in the machine that SAS is using.

9.02 Poll
Have you ever run out of space during a sort?
€ Yes
€ No

17
9-8 Chapter 9 Sorting SAS Data Sets

Sort Space Requirements

proc sort data=orion.orders;
by Order_Date Order_ID;
run;

orion
Disk Space
orders
orders

SORT utility work space

Space required for the SAS sort
18 p309d01

Sort Space Requirements

The amount of space that the SAS sort needs depends
on the following conditions:
„ whether the sort can be done with threading

„ the length of the observations

„ the number of variables in the BY statement and their

storage lengths
„ the operating environment in which PROC SORT
executes
„ the library to which the sorted data is written

 A quick rule-of-thumb method for estimating the

space requirements for sorting with the SAS sort is
four times the size of the SAS data set being sorted.

19
 9.1 Using the SORT Procedure 9-9

Reference Information

The formula below calculates the estimated amount of space needed by a single-threaded PROC SORT:
bytes required=((4 * obslen) + (2 * keylen)) * numobs
The formula below calculates the estimated amount of space needed by a multi-threaded PROC SORT:
bytes required=3 * (obslen * numobs)
The space calculation for the SAS 8.2 sort is as follows:
bytes required=(keylen + obslen) * numobs * N

obslen length of the observation

keylen length of the BY variables when concatenated to form a single value

numobs number of observations in the data set

Use the CONTENTS or DATASETS procedure to gather the required information.

proc contents data=orion.orders;
run;
These space calculations assume that the SAS®9 sort can take place in memory without using utility swap
files.
If you use the OVERWRITE option in the PROC SORT statement in multi-threaded environments, you
need space equal to the data set size. The OVERWRITE option enables the input data set to be deleted
before the replacement output data set is populated with observations. The OVERWRITE option is
supported by the SAS sort and SAS multi-threaded sort only. The option has no effect if you use a host
sort or the TAGSORT option.
Use the OVERWRITE option only with a data set that is backed up or with a data set that you can
reconstruct. Because the input data set is deleted, data will be lost if a failure occurs while the output data
set is being written.
9-10 Chapter 9 Sorting SAS Data Sets

Using the SORTSIZE= Option

Use the SORTSIZE= option in the PROC SORT
statement to do the following:
„ specify the amount of memory that is available to the
SORT procedure
„ improve the sort performance by restricting the
swapping of memory to disk that is controlled by the
operating system
General form of the SORTSIZE= option:

SORTSIZE=n | nK | nM | nG | MIN | MAX | hexX | SIZE;

20

Allocating Sort Workspace

Actual Required Sorting Processing
Workspace Time
less than or equal occurs in memory reduced
to SORTSIZE=
SAS system option
greater than utility files on disk increased
SORTSIZE= SAS and memory
system option

The multi-threaded SAS®9 sort fails to complete

a sort if the value of SORTSIZE= is too small.

21
 9.1 Using the SORT Procedure 9-11

The SORTSIZE= Option

For optimal performance, set the SORTSIZE= option to
a value less than the available physical memory. This
enables the programs and the operating environment
to stay resident in memory.
You should investigate how resources are affected
if you change the value of the SORTSIZE= option.

options fullstimer;
proc sort data=orion.order_fact
sortsize=max;
by Order_Date;
run;

p309d02
22

9.03 Multiple Choice Poll

Do you have a host sort utility such as SYNCSORT or
DFSORT available?
€ Yes
€ No

24
9-12 Chapter 9 Sorting SAS Data Sets

Host Sort Utilities

A host sort utility has the following characteristics:
„ is a third-party sort package

„ is available for UNIX, z/OS, and Windows platforms

 Ask your system administrator whether a host sort

utility is available at your site.

25

SAS System Options for Selecting a Host Sort

SAS System What It Does Syntax
Option
SORTPGM= specifies which sort OPTIONS
utility to use SORTPGM=utility | BEST |
HOST | SAS;

SORTCUTP= specifies a cut-over OPTIONS SORTCUTP=n |

point in terms of the nK | nM | nG | MAX | MIN |
size (in bytes) of a hexX;
SAS data set
SORTNAME= specifies the host sort OPTIONS
utility to be invoked if SORTNAME=host-sort-
SORTPGM=BEST | utility-name;
HOST

26
 9.1 Using the SORT Procedure 9-13

Sorted Data
When data is sorted by SAS, the descriptor contains the
following information:
1. a sort indicator that contains the variable(s) on which
the data is sorted
2. whether the sort is validated by SAS
3. the character set used
Additional information contained in the descriptor portion
includes the following:
4. the collating sequence used for ordering the data
5. collation rules, if the data set is sorted linguistically

27

Sorted Data
Partial PROC CONTENTS Output
The CONTENTS Procedure

Data Set Name ORION.SALESSTAFF Observations 163

Member Type DATA Variables 10
Engine V9 Indexes 0
Created Monday, December 17, 2007 09:04:37 PM Observation Length 136
Last Modified Monday, December 17, 2007 09:04:37 PM Deleted Observations 0
Protection Compressed NO
Data Set Type Sorted YES
Label
Data Representation WINDOWS_32
Encoding wlatin1 Western (Windows)

<lines removed >

Sort Information

Sortedby Emp_Hire_Date
Validated YES
Character Set ANSI

p309d03
28

ANSI (American National Standards Institute) is an organization in the United States that coordinates
voluntary standards and conformity to those standards. ANSI works with the ISO (International
Organization for Standardization) to establish global standards.
By default, the ANSI character set uses ASCII (American Standard Code for Information Interchange) for
the Windows and UNIX operating environments and EBCDIC (Extended Binary Coded Decimal
Interchange Code) for z/OS.
9-14 Chapter 9 Sorting SAS Data Sets

9.04 Multiple Answer Poll

If you sort a SAS data set that is in sorted order, does
SAS re-sort the data?
a. No, if the sort indicator is 'YES', and the validation
is 'YES'
b. No, if the sort indicator is 'YES', and the validation
is 'NO'
c. Yes, regardless of the sort indicator or validation
d. Yes, if the sort indicator is 'YES', and the validation
is 'NO'

30

Setting Sort Indicator and Validation

The sort indicator and validation can be set by one of the
following techniques:
„ sorting the data using PROC SORT or PROC SQL
with an ORDER BY clause
„ specifying the sort order by using the SORTEDBY=
data set option
„ using the PRESORTED option in the PROC SORT
statement

32
 9.1 Using the SORT Procedure 9-15

Using the SORTEDBY= Option

If the input data is in sorted order, you can specify the
order by using the SORTEDBY= output data set option.
The SORTEDBY= option has the following attributes:
„ sets the sort indicator on the data set to YES

„ defines the sort indicator as an asserted data order

(not validated)
„ requires that SAS check the order of the data
as it processes it
General form of the SORTEDBY option:

data-set-name(SORTEDBY=BY-clause | _NULL_ )

33

BY-clause indicates the data order.

_NULL_ removes any existing sort information.

Using the SORTEDBY= Option

Create a SAS data set from an external file containing
invoice information. The external file is in sorted order
by Order_Date.
filename M1 'mon1.dat';

data january(sortedby=Order_Date);
infile M1 dlm=',';
input Customer_ID Order_ID Order_Type
Order_Date : date9.
Delivery_Date : date9.;
run;

p309d04
34
9-16 Chapter 9 Sorting SAS Data Sets

Using the SORTEDBY= Option

Partial SAS Log
The CONTENTS Procedure

Data Set Name WORK.JANUARY Observations 4

Member Type DATA Variables 5
Engine V9 Indexes 0
Created Sunday, January 27, 2008 05:36:23 PM Observation Length 40
Last Modified Sunday, January 27, 2008 05:36:23 PM Deleted Observations 0
Protection Compressed NO
Data Set Type Sorted YES
Label
Data Representation WINDOWS_32
Encoding wlatin1 Western (Windows)

<lines removed>

Sort Information

Sortedby Order_Date
Validated NO
Character Set ANSI

p309d04
35

Using the SORTEDBY= Option

Attempt to sort the data.
proc sort data=january;
by Order_Date;
run;

Log
1197 proc sort data=january;
1198 by Order_Date;
1199 run;

NOTE: Input data set is already sorted, no sorting done.

NOTE: PROCEDURE SORT used (Total process time):
real time 0.03 seconds
cpu time 0.00 seconds

p309d05
36
 9.1 Using the SORT Procedure 9-17

Using the PRESORTED Option

Beginning in SAS 9.2, there is a PROC SORT statement
option, PRESORTED, that checks within the input data
set to determine whether the sequence of observations is
in order before sorting. By specifying this option, you can
avoid the cost of sorting the data set.

proc sort data=orion.salesstaff presorted;

by Emp_Hire_Date;
run;

 If the data set orion.salesstaff is not in sorted

order by Emp_Hire_Date, PROC SORT with the
PRESORTED option sorts the data.

p309d06
37

In SAS 9.2, the SORTVALIDATE system option specifies whether the SORT procedure verifies that a
data set is sorted according to the variables in the BY statement when the sort indicator metadata
designates a user-specified sort order. NOSORTVALIDATE is the default.

OPTIONS NOSORTVALIDATE | SORTVALIDATE;

Using the PRESORTED Option

Partial Log
1 proc sort data=orion.salesstaff presorted;
2 by Emp_Hire_Date;
3 run;

NOTE: Input data set is already sorted, no sorting done.

NOTE: PROCEDURE SORT used (Total process time):

Partial PROC CONTENTS Output

Sort Information

Sortedby Emp_Hire_Date
Validated YES
Character Set ANSI

p309d06
38
9-18 Chapter 9 Sorting SAS Data Sets

Setup for the Poll

p905d04
filename M1 'mon1.dat';
data january(sortedby=Order_Date);
infile M1 dlm=',';
input Customer_ID Order_ID Order_Type
Order_Date : date9.
Delivery_Date : date9.;
run;
proc contents data=january;
run;

p905d05
proc sort data=january;
by Order_Date;
run;

40

9.05 Multiple Choice Poll

What would be the effect of using the PRESORTED
option in the SORT procedure step in p309d05?
a. The data set January would be sorted and the
sort validation flag set to 'YES'.
b. The data set January would not be sorted and the
sort validation flag set to 'YES'.
c. The data set January would not be sorted and the
sort validation flag set to 'NO'.

41
 9.1 Using the SORT Procedure 9-19

Sort Order
The character set determines the sort order of a particular
character in relation to other characters. By default,
PROC SORT uses one of the following collating
sequences, depending on the environment under which
the procedure is running:
„ ASCII (Windows and UNIX)

„ EBCDIC (z/OS)

 In addition, the EQUALS|NOEQUALS option in the

PROC SORT statement specifies the order of the
observations within a BY group in the output
data set.

43

Collating Default Sort Order

Sequence

ASCII blank ! " # $ % & ' ( ) * + , - . /0 1 2 3 4 5 6 7 8 9 : ; < = > ? @ A

B C D E F G H I J K L M N O P Q R S T U V W X Y Z[ \] ˆ_ a b
cdefghijklmnopqrstuvwxyz{}~

EBCDIC blank . < ( + | & ! $ * ) ; ¬ - / , % _ > ?: # @ ' = " a b c d e f g h i j

klmnopqr~stuvwxy z{ABCDEFGHI}JKLMN
O P Q R \S T U V W X Y Z 0 1 2 3 4 5 6 7 8 9

The options EBCDIC, ASCII, NATIONAL, DANISH, SWEDISH, and REVERSE can change the default
collating sequence.
9-20 Chapter 9 Sorting SAS Data Sets

Using the EQUALS | NOEQUALS Option

„ For observations with identical BY-variable values,
EQUALS maintains the relative order of the
observations within the input data set in the output
data set. This is the default.
„ NOEQUALS does not necessarily preserve this order
in the output data set.

PROC SORT DATA=SAS-data-set EQUALS | NOEQUALS ;

 Using NOEQUALS can save CPU and memory.

44

9.06 Quiz
The data set orion.salesstaff was previously sorted in
ascending order by Employee_Hire_Date.
Open and submit program p309a02.
proc sort data=orion.salesstaff
out=salesstaff;
by Employee_ID;
run;
proc print data=salesstaff;
where Employee_ID=120134;
run;

For employee number=120134, which date is first?

a. The first date hired
b. The last date hired

46
 9.1 Using the SORT Procedure 9-21

Using the EQUALS and NOEQUALS Options

This demonstration illustrates using the EQUALS and NOEQUAL options.

p309d07
proc sort data=orion.salesstaff
out=salesstaff_eq equals;
by Employee_ID;
run;
proc sort data=orion.salesstaff
out=salesstaff_noeq noequals;
by Employee_ID;
run;
proc compare data=salesstaff_eq compare=salesstaff_noeq;
run;
Partial Output
The COMPARE Procedure

Comparison of WORK.SALESSTAFF_EQ with WORK.SALESSTAFF_NOEQ

(Method=EXACT)

Data Set Summary

Dataset Created Modified NVar NObs

WORK.SALESSTAFF_EQ 15JUN10:07:23:01 15JUN10:07:23:01 10 163

WORK.SALESSTAFF_NOEQ 15JUN10:07:23:01 15JUN10:07:23:01 10 163

Variables Summary

Number of Variables in Common: 10.

Observation Summary

Observation Base Compare

First Obs 1 1
First Unequal 14 14
Last Unequal 49 49
Last Obs 163 163

Number of Observations in Common: 163.

Total Number of Observations Read from WORK.SALESSTAFF_EQ: 163.
Total Number of Observations Read from WORK.SALESSTAFF_NOEQ: 163.

Number of Observations with Some Compared Variables Unequal: 4.

Number of Observations with All Compared Variables Equal: 159.

Values Comparison Summary

Number of Variables Compared with All Observations Equal: 6.

Number of Variables Compared with Some Observations Unequal: 4.
Number of Variables with Missing Value Differences: 1.
Total Number of Values which Compare Unequal: 14.
Maximum Difference: 12387.
9-22 Chapter 9 Sorting SAS Data Sets

Using the NODUPKEY and DUPOUT= Options

When you use the NODUPKEY option to remove
observations in the output data set, the choice of
EQUALS or NOEQUALS can affect which observations
are removed.
The SORT procedure NODUPKEY and DUPOUT=
options used together create the following output data
sets:
„ oneemp containing exactly one observation per
Employee_ID
„ extra containing the four duplicate Employee_ID
observations
proc sort data=orion.salesstaff nodupkey
out=oneemp
dupout=extra;
by Employee_ID;
run;
49 p309d08

Using the NODUPKEY and DUPOUT= Options

General form of the DUPOUT= and NODUPKEY options:
PROC SORT DATA=data-set-name
OUT=data-set-name-2
DUPOUT=data-set-name-3
NODUPKEY;
BY BY-variable(s);
RUN;

DUPOUT= specifies the output data set to which

duplicate observations are written.

NODUPKEY checks for and eliminates observations

with duplicate BY values.

50
 9.1 Using the SORT Procedure 9-23

Changing the Collating Sequence

You can change the collating sequence by specifying one
of the these options in the PROC SORT statement.
„ ASCII

„ EBCDIC

„ DANISH

„ FINNISH

„ NORWEGIAN

„ POLISH

„ SWEDISH

„ NATIONAL

„ SORTSEQ=

You can specify only one collating sequence option

in a PROC SORT step.
51

 Refer to the “Collating Sequence” chapter of the SAS National Language Support (NLS): User’s
Guide for detailed information about the various collating sequences and when they are used.

Setup for the Poll

The data set orion.customer contains a variable named
Customer_Address.
Partial Listing of orion.customer (Selected Observations)
Customer_ID Country . . . Customer_Address . . .
. . . . . .
29 AU 21 Hotham Parade
41 AU . . . 9 Angourie Court . . .

63 US . . . 25 Briarforest Pl . . .

111 AU . . . 28 Munibung Road . . .

21 Parliament House c/-

171 AU . . . . . .
Senator t
183 AU . . . 18 Fletcher Rd . . .

4 Burke Street
195 AU . . . . . .
Woolloongabba
215 AU . .. 23 Benjamin Street . ..

46966 CA . . . 17 boul Wallberg . . .

70221 CA . . . 9 South Service Rd . . .

53
9-24 Chapter 9 Sorting SAS Data Sets

9.07 Multiple Choice Poll

If the data set orion.customer is sorted by
Customer_Address, which of the following values would
precede the other value in the new ordered data set?
a. 21 Hotham Parade
b. 9 Angourie Court

54

Using the LINGUISTIC Option

In SAS 9.2, the LINGUISTIC value for the SORTSEQ=
option in the PROC SORT statement specifies that the
rules and default collating sequence options are based on
the language specified in the current value of the SAS
system option LOCALE.
For example, the data set orion.customer contains
customers from seven countries. The countries and
associated locales are listed below:
Country Locale
Australia en_AU
Canada en_CA
Germany de_DE
Israel he_IL
Turkey tr_TR
United States en_US
56 South Africa en_ZA
 9.1 Using the SORT Procedure 9-25

Using the LINGUISTIC Suboptions

The following are selected collation rules that can be
specified for the LINGUISTIC option. These rules modify
the linguistic collating sequence.
Suboption Values Purpose
CASE_FIRST= UPPER | LOWER Either value can sort data so
that like letters are sorted
correctly irrespective of case.
NUMERIC_COLLATION= OFF | ON This orders integer values
within the text by the numeric
value instead of characters
used to represent the
numbers.
STRENGTH= PRIMARY | The PRIMARY value supports
SECONDARY | sorting that is not case
TERTIARY | sensitive. The other values
QUATERNARY | handle diacritical and
IDENTICAL punctuation differences.
57

The following table contains the values of the CASE_FIRST= suboption:

UPPER sorts uppercase letters first, and then the lowercase letters.

LOWER sorts lowercase letters first, and then the uppercase letters.
9-26 Chapter 9 Sorting SAS Data Sets

The following table contains the values of the STRENGTH= suboption:

Value Type of Collation Description

PRIMARY or 1 PRIMARY specifies differences It is the strongest difference. For
between base characters (for example, dictionaries are divided into
example, "a" < "b"). different sections by base character.
SECONDARY or 2 Accents in the characters are A secondary difference is ignored when
considered secondary differences there is a primary difference anywhere in
(for example, "as" < "às" < "at"). the strings. Other differences between
letters can also be considered secondary
differences, depending on the language.
TERTIARY or 3 Uppercase and lowercase differences A tertiary difference is ignored when
in characters are distinguished at the there is a primary or secondary
tertiary level (for example, "ao" < difference anywhere in the strings.
"Ao" < "aò"). Another example is the difference
between large and small Kana.
QUATERNARY or 4 When punctuation is ignored at level The quaternary level should be used if
1-3, an additional level can be used ignoring punctuation is required or when
to distinguish words with and processing Japanese text. This difference
without punctuation (for example, is ignored when there is a primary,
"ab" < "a-b" < "aB"). secondary or tertiary difference.
IDENTICAL or 5 When all other levels are equal, the This level should be used sparingly, as
identical level is used as a tiebreaker. only code point values differences
The Unicode code point values of between two strings is an extremely rare
the Normalization Form D (NFD) occurrence. For example, only Hebrew
form of each string are compared at cantillation marks are distinguished at
this level, in case there is no this level.
difference at levels 1-4.
 9.1 Using the SORT Procedure 9-27

Using the NUMERIC_COLLATION= Option

This demonstration illustrates sorting data using the NUMERIC_COLLATION=ON option.

p309d09
proc sort data=orion.customer
out=customer
sortseq=linguistic (numeric_collation=on);
by Customer_Address;
run;

proc contents data=customer;

title 'Using the SORTSEQ= Option';
run;

proc print data=customer;

var Customer_ID Customer_Address Country;
run;
Partial PROC CONTENTS Output
Using the SORTSEQ= Option

The CONTENTS Procedure

Data Set Name WORK.CUSTOMER Observations 77

Member Type DATA Variables 12
Engine V9 Indexes 0
Created Saturday, June 26, 2010 07:52:03 AM Observation Length 200
Last Modified Saturday, June 26, 2010 07:52:03 AM Deleted Observations 0
Protection Compressed NO
Data Set Type Sorted YES
Label
Data Representation WINDOWS_32
Encoding wlatin1 Western (Windows)

< lines removed >

Sort Information

Sortedby Customer_Address
Validated YES
Character Set ANSI
Collating Sequence LINGUISTIC

Sort Information

Locale en_US
Strength 3
Numeric Collation ON
9-28 Chapter 9 Sorting SAS Data Sets

PROC PRINT Output

Using the SORTSEQ= Option

Obs Customer_ID Customer_Address Country

1 195 4 Burke Street Woolloongabba AU

2 41 9 Angourie Court AU
3 70221 9 South Service Rd CA
4 46966 17 boul Wallberg CA
5 183 18 Fletcher Rd AU
6 29 21 Hotham Parade AU
7 171 21 Parliament House c/- Senator t AU
8 215 23 Benjamin Street AU
9 63 25 Briarforest Pl US
10 111 28 Munibung Road AU
11 70210 40 Route 199 CA
12 26148 41 Main St CA
13 31 42 Arrowood Ln US
14 11171 69 chemin Martin CA
15 75 101 Knoll Ridge Ln US
16 18 117 Langtree Ln US
17 53 131 Franklin St AU
18 49 185 Birchford Ct US
19 27 188 Grassy Creek Pl US
20 90 252 Clay St US
21 71 290 Glenwood Ave US
22 70079 304 Grand Lake Rd CA
23 70201 319 122 Ave NW CA
24 56 334 Kingsmill Rd US
25 17 391 Greywood Dr US
26 36 417 Halstead Cir US
27 92 421 Blue Horizon Dr US
28 10 425 Bryant Estates Dr US
29 24 435 Cambrian Way US
30 54655 512 Gregoire Dr CA
31 70059 580 Howe St CA
32 70100 614 Route 199 CA
33 70046 818 rue Davis CA
34 17023 837 rue Lajeunesse CA
35 34 844 Glen Eden Dr US
36 70165 873 rue Bosse CA
37 70108 1001 Burrard St CA
38 12 1068 Haithcock Rd US
39 52 1233 Hunters Crossing US
40 23 1532 Ferdilah Ln US
41 70187 1835 boul Laure CA
42 20 2187 Draycroft Pl US
43 60 2429 Hunt Farms Ln US
44 89 2572 Glenharden Dr US
45 88 3815 Askham Dr US
46 4 4382 Gralyn Rd US
47 39 4386 Hamrick Dr US
48 69 4948 Dargan Hills Dr US
49 5 6468 Cog Hill Ct US
50 45 7818 Angier Rd US

(Continued on the next page.)

 9.1 Using the SORT Procedure 9-29

Using the SORTSEQ= Option

Obs Customer_ID Customer_Address Country

51 79 9658 Dinwiddie Ct US
52 544 A Blok No: 1 TR
53 1100 A Blok No: 1 TR
54 1684 A Blok No: 1 TR
55 2618 Arnold Road 2 ZA
56 65 Bahnweg 1 DE
57 2550 Bryanston Drive 122 ZA
58 42 Carl Von Linde Str. 13 DE
59 11 Carl-Zeiss-Str. 15 DE
60 1033 Fahrettin Kerim Gokay Cad. No. 24 TR
61 2788 Fahrettin Kerim Gokay Cad. No. 30 TR
62 19 Hechtsheimerstr. 18 DE
63 50 Humboldtstr. 1 DE
64 13 Iese 1 DE
65 9 Kallstadterstr. 9 DE
66 908 Mayis Cad. Nova Baran Plaza Ka 11 TR
67 14703 Mivtza Boulevard 17 IL
68 12386 Mivtza Kadesh St 16 IL
69 19873 Mivtza Kadesh St 18 IL
70 14104 Mivtza Kadesh St 25 IL
71 19444 Mivtza Kadesh St 61 IL
72 3959 Moerbei Avenue 120 ZA
73 33 Münsterstraße 67 DE
74 61 Münzstr. 28 DE
75 16 Oberstr. 61 DE
76 2806 Quinn Street 11 ZA
77 928 Turkcell Plaza Mesrutiyet Cad. 142 TR
9-30 Chapter 9 Sorting SAS Data Sets

Exercises

Level 1

1. Using the PRESORTED Option

a. Use PROC CONTENTS to determine whether the data set orion.holidays is sorted.
b. Write a PROC SORT step to sort the data orion.holidays by Date. Create a temporary data set
named holidays. Use the PRESORTED option in the PROC SORT statement.
What is the resulting message in the log?

c. Submit a PROC CONTENTS step to determine whether the data set holidays is sorted by Date.
d. Change the BY variable to Holiday_Name and resubmit the PROC SORT step.
What is the resulting message in the log?

e. Submit a PROC CONTENTS step to determine whether the data set holidays is sorted by
Holiday_Name.

Level 2

2. Creating a Sorted Data Set

a. Open the program p309e02.
b. Modify the program to create a data set named profit07 and specify that it is sorted by Company
without sorting the data set.
p309e02
data profit07;
infile 'profit07.dat' dlm=',';
input Company:$30. Sales Cost Salaries Profit;
run;
c. Use PROC CONTENTS to verify that profit07 has a sort flag on the variable Company.
 9.1 Using the SORT Procedure 9-31

d. Use PROC PRINT to create a report grouped by the Company variable. Print the first 24
observations of the data set.
Partial PROC PRINT Output
------------------------- Company=Logistics --------------------------

Obs Sales Cost Salaries Profit

1 124918 535949 247341 163690

2 127416 389429 181185 80829
3 131901 313762 145103 36757
4 134182 642970 309099 199690
< lines omitted >
9 144364 241863 113212 -15713
10 144364 339880 158696 36821
11 150958 432257 199526 81773
12 150958 947060 442396 353706

---------------------- Company=Orion Australia -----------------------

Obs Sales Cost Salaries Profit

13 119597 159404 74771 -34964

14 123908 146004 69396 -47300
15 132940 64236 30403 -99107
< lines omitted >
21 162870 79060 36898 -120708
22 181463 163110 77705 -96057
23 188318 186131 87454 -89641
24 212579 257931 121060 -75708

e. Use the PROC SORT PRESORTED option to turn on the validated flag and use
PROC CONTENTS to verify that the sort flag is set in the descriptor portion.
9-32 Chapter 9 Sorting SAS Data Sets

Level 3

3. Using the SORTSEQ=LINGUISTIC Option

The first five observations of orion.spain_customers are listed below:
Personal_ Customer_
Obs Customer_ID Country Gender ID Customer_Name FirstName

1 2 ES F Mercedes Martínez Mercedes

2 7 ES F Julián Escorihuela Monserrate Julián
3 21 ES M José Fernández de Mesa José
4 72 ES F Mónica Aranda Unzurrunzaga Mónica
5 73 ES F Pilar Bachs Pallarés Pilar

Birth_ Street_ Customer_

Obs Customer_LastName Date Customer_Address Street_ID Number Type_ID

1 Martínez 15JAN1955 Edificio 2 8300101034 2 1010

2 Escorihuela Monserrate 07AUG1975 Co. De Los Claveles 561 8300100945 561 1040
3 Fernández de Mesa 16JUN1955 C. Santa Hortensia 2 8300100682 2 2010
4 Aranda Unzurrunzaga 04JAN1975 Apartado 64 8300100254 64 1020
5 Bachs Pallarés 04JAN1945 C. Conde de Peñalver 68 8300100485 68 1030

a. Open and submit the program p309e03.

What is the value of the LOCALE= option?
p309e03
proc options option=locale;
run;
b. Write a PROC SORT step to sort orion.spain_customers by Customer_Name.
c. Create an output data set named s_customers.
d. Use the SORTSEQ=LINGUISTIC option with the LOCALE=es_ES suboption.
e. Print the data set s_customers and look at the order of the variable Customer_Name.
 9.2 BY-Group Processing (Self-Study) 9-33

9.2 BY-Group Processing (Self-Study)

Objectives
„ Define BY-group processing.
„ Use indexes to return the data in sorted order.
„ Use indexes to combine data horizontally.
„ Use a format to group data for BY-group processing.
„ Use a CLASS statement.
„ Specify a user-asserted sort order.

62

BY-Group Processing
BY-group processing has these characteristics:
„ is a method of processing observations that are
grouped or ordered by the values of the BY variables
„ can be used in both DATA and PROC steps

„ can be used to eliminate observations with duplicate

BY values
The following techniques can be used to perform BY-
group processing:
„ the SORT procedure

„ indexes on the data set

„ the NOTSORTED option in the BY statement

„ user-sort assertion

„ a CLASS statement
63
9-34 Chapter 9 Sorting SAS Data Sets

9.08 Multiple Choice Poll

Which of the following would be a good use of an index
for BY-group processing?
a. Your data set is sorted by the BY variable already.
b. Your data set is updated using a SET statement in the
DATA step. In that case, the index would be updated
after the append, and you would not have to reindex
the data.
c. Your data is updated using PROC APPEND, PROC
SQL, or the MODIFY statement in the DATA step. In
that case, the index would be updated after the
update, and you would not have to reindex the data.

65

Using an Index for BY-Group Processing

BY-group processing with an index eliminates the need to
sort data.
„ Having multiple indexes enables sequencing data by
different variables without having to repeat the SORT
procedure.
„ Indexes are updated when observations are modified
or added to a SAS data set.
options msglevel=i;
proc print data=orion.retail(obs=10);
title 'Retail Sales';
by Customer_ID;
var Customer_ID Product_ID Quantity
Total_Retail_Price;
run;

67 p309d10
 9.2 BY-Group Processing (Self-Study) 9-35

Limitations of Using an Index for BY-Group

Processing
BY-group processing with an index has the following
limitations:
„ less efficient than sequentially reading a sorted data
set
„ extra memory requirement to use the index

„ extreme I/O increase

Partial SAS Log

608 options msglevel=i;
609 proc print data=orion.retail(obs=10);
610 title 'Retail Sales';
611 by Customer_ID;
INFO: Index Customer_ID selected for BY clause processing.
NOTE: An index was selected to execute the BY statement.
The observations will be returned in index order rather than in physical
order. The selected index is for the variable(s):
Customer_ID
612 var Customer_ID Product_ID Quantity Total_Retail_Price;
68 613 run; p309d10

Using an Index for BY-Group Processing

Partial PROC PRINT Output
Retail Sales

-------------------------- Customer ID=4 --------------------------

Total_Retail_
Obs Customer_ID Product_ID Quantity Price

62 4 240600100017 1 $53.00
70 4 220101400145 1 $16.70
79 4 240700100011 3 $80.97
111 4 230100100053 2 $92.60

------------------------- Customer ID=5 --------------------------

Total_Retail_
Obs Customer_ID Product_ID Quantity Price

37 5 240100100433 1 $3.00
48 5 220101400276 2 $136.80
88 5 240300200018 1 $87.20
89 5 240300300071 1 $138.00
148 5 220101400265 2 $74.20
149 5 220101400387 4 $50.40

69
9-36 Chapter 9 Sorting SAS Data Sets

When a BY Statement Does Not Use an Index

A BY statement does not use an index if either of the
following conditions are present:
„ The BY statement includes the DESCENDING
or NOTSORTED option.
„ SAS is aware that the data file is physically stored
in sorted order according to the BY variables.

70

Business Scenario
The data set orion.street_code contains the Street_ID
and the name, city, and country for the streets.

Partial Listing of orion.street_code

Street_ Postal_
Street_ID Country City_Name
Name Code
1600100760 AU Fletcher Rd 4170 Colnslie
1600101527 AU Angourie Court 5009 Kilkenny
Kingston-
1600101555 AU Parliament Road 5331
On-Murray
Burke Street
1600101663 AU 217 Liverpool
Woolloongabba
1600101794 AU Brunswick Street 8008 Melbourne

71
 9.2 BY-Group Processing (Self-Study) 9-37

Business Scenario
The SAS data set orion.order_fact contains the
information needed for the delivery of products to Orion
customers. There is no index on Street_ID in the
orion.order_fact data set.
Partial Listing of orion.order_fact
Customer_
Street_ID Delivery_Date Order_ID Quantity
ID
63 9260125492 11JAN2003 1230058123 1
5 9260114570 19JAN2003 1230080101 1
45 9260104847 22JAN2003 1230106883 1
41 1600101527 28JAN2003 1230147441 2
183 1600100760 27FEB2003 1230315085 3
79 9260101874 03MAR2003 1230333319 1
23 9260126679 08MAR2003 1230338566 1
23 9260126679 08MAR2003 1230338566 2
45 9260104847 11MAR2003 1230371142 2
72

Business Scenario
Combine the two data sets to expedite delivery of
customer orders.
1. Create an index on Street_ID in orion.street_code.
2. Use SET/SET with KEY=Street_ID to combine the data
sets.
Partial Listing of addresses
Street_ Postal_ City_
Customer_ID Street_ID . . . Name Code Name
Briarforest
63 9260125492 . . . Pl
62201 St. Clair

5 9260114570 . . . Cog Hill Ct 90280 Los Angeles

45 9260104847 . . . Angier Rd 94520 Contra Costa

Angourie
41 1600101527 . . . Court
5009 Kilkenny

183 1600100760 . . . Fletcher Rd 4170 Colnslie

79 9260101874 . . . Dinwiddie Ct 16648 Blair

73
9-38 Chapter 9 Sorting SAS Data Sets

Setup for the Poll

p309a03
proc datasets library=orion nolist;
modify street_code;
index create Street_ID;
quit;

data addresses;
set orion.order_fact(keep=Customer_ID Street_ID
Delivery_Date Order_ID Product_ID Quantity);
set orion.street_code(keep=Street_ID Country Street_Name
City_Name Postal_Code)
key=Street_ID / unique;
run;

proc print data=addresses;

title 'Customer Addresses';
run;

75

9.09 Multiple Choice Poll

How is orion.order_fact read?
a. Direct access via the index
b. Sequentially

76
 9.2 BY-Group Processing (Self-Study) 9-39

9.10 Multiple Choice Poll

How is orion.street_code read?
a. Direct access via the index
b. Sequentially

78

Using Indexes
proc datasets library=orion nolist;
modify street_code;
index create Street_ID;
orion.order_fact
quit; is read sequentially.

data addresses;
set orion.order_fact(keep=Customer_ID
Street_ID
Delivery_Date
Order_ID
Product_ID
Quantity) ;
set orion.street_code(keep=Street_ID
Country
Street_Name
City_Name
orion.street_code is read
Postal_Code)
key=Street_ID / unique; by accessing the index and
run;
directly accessing the
appropriate observation.

p309a03
80

 The UNIQUE option causes a KEY= search to use the first matching observation from the
indexed data set, if there are duplicates.
9-40 Chapter 9 Sorting SAS Data Sets

The UNIQUE Option

Option Processing Results
none begins searching at the Run-time error occurs if
top of the index only when there are fewer duplicates in
the KEY= value changes the indexed data. The PDV
(sequential matching contains values from the
operation on consecutive last match.
duplicate key values).
UNIQUE always begins searching Duplicate values in the first
at the top of the index. data set are always
matched with the first
matching key value in the
indexed data set.

UNIQUE does not indicate that the keys are

unique in the indexed data set.
81

Duplicate Key Values (Review)

data three;
set one;
set two key=Variable;
run;

Example 1: Contiguous duplications in one

one two
Variable Variable
A A
A B
A C
No match
Run-time error
82
 9.2 BY-Group Processing (Self-Study) 9-41

Duplicate Key Values (Review)

data three;
set one;
set two key=Variable / unique;
run;

Example 2: Contiguous duplications in one with the

one UNIQUE option two
Variable Variable
A A
A B
A C

83

9.11 Quiz
Open and submit the program p309a03.
Are there any data errors in the log?
data addresses;
set orion.order_fact(keep=Customer_ID Street_ID
Delivery_Date Order_ID Product_ID Quantity);
set orion.street_code(keep=Street_ID Country
Street_Name City_Name Postal_Code)
key=Street_ID / unique;
run;

85
9-42 Chapter 9 Sorting SAS Data Sets

9.12 Quiz
Open and submit the program p309a04.
1. What does the SAS log show for this DATA step?
2. Why do you get those error messages?

data addresses2;
set orion.order_fact(keep=Customer_ID Street_ID
Delivery_Date Order_ID Product_ID Quantity);
set orion.street_code(keep=Street_ID Country
Street_Name City_Name Postal_Code)
key=Street_ID;
run;

87

Comparing the Data Sets

proc compare data=addresses
compare=addresses2;
run;

The data values in the addresses data set and the data
values in the addresses2 data sets are equal.
Observation Summary

Observation Base Compare

First Obs 1 1
Last Obs 617 617

Number of Observations in Common: 617.

Total Number of Observations Read from WORK.ADDRESSES: 617.
Total Number of Observations Read from WORK.ADDRESSES2: 617.

Number of Observations with Some Compared Variables Unequal: 0.

Number of Observations with All Compared Variables Equal: 617.

NOTE: No unequal values were found. All values compared are exactly equal.
p309a04
90
 9.2 BY-Group Processing (Self-Study) 9-43

Business Scenario
You must print the data set orion.shoe_vendors
by Group_Name for the vendors where the variable
Mfg_Suggested_Retail_Price is greater than $100.
Mfg_Suggested_Retail_Price>100

Mfg_Suggested_
Obs Group_Name Supplier_Name Category_Name Retail_Price

4 Eclipse, Kid's Shoes Eclipse Inc Children Sports $114.00

13 Eclipse, Kid's Shoes Eclipse Inc Children Sports $130.00
29 Eclipse, Kid's Shoes Eclipse Inc Children Sports $114.00
30 Eclipse, Kid's Shoes Eclipse Inc Children Sports $117.00
43 Tracker Kid's Clothes 3Top Sports Children Sports $106.00
55 Tracker Kid's Clothes 3Top Sports Children Sports $113.00
59 Tracker Kid's Clothes 3Top Sports Children Sports $109.00
68 LSF Eclipse Inc Clothes $141.00
69 LSF Petterson AB Clothes $130.00
70 LSF Petterson AB Clothes $140.00
71 LSF 3Top Sports Clothes $144.00
72 LSF Petterson AB Clothes $125.00
73 LSF Triple Sportswear Inc Clothes $162.00

91

Setup for the Poll

Partial Listing
Mfg_Suggested_Retail_Price>100

Mfg_Suggested_
Obs Group_Name Supplier_Name Category_Name Retail_Price

4 Eclipse, Kid's Shoes Eclipse Inc Children Sports $114.00

13 Eclipse, Kid's Shoes Eclipse Inc Children Sports $130.00
29 Eclipse, Kid's Shoes Eclipse Inc Children Sports $114.00
30 Eclipse, Kid's Shoes Eclipse Inc Children Sports $117.00
43 Tracker Kid's Clothes 3Top Sports Children Sports $106.00
55 Tracker Kid's Clothes 3Top Sports Children Sports $113.00
59 Tracker Kid's Clothes 3Top Sports Children Sports $109.00
68 LSF Eclipse Inc Clothes $141.00
69 LSF Petterson AB Clothes $130.00
70 LSF Petterson AB Clothes $140.00
71 LSF 3Top Sports Clothes $144.00
72 LSF Petterson AB Clothes $125.00
73 LSF Triple Sportswear Inc Clothes $162.00

proc print data=orion.shoe_vendors;

title 'Mfg_Suggested_Retail_Price>100';
where Mfg_Suggested_Retail_Price>100;
var Group_Name Supplier_Name Category_Name
Mfg_Suggested_Retail_Price;
run;
93 p309a05
9-44 Chapter 9 Sorting SAS Data Sets

9.13 Multiple Choice Poll

Which of the following is true of the data set
orion.shoe_vendors?
a. The data set is sorted by the variable Group_Name.
b. The data set is not sorted by the variable
Group_Name. However, it is grouped by
Group_Name.
c. The data set is neither sorted nor grouped by the
variable Group_Name.

94

Using the NOTSORTED Option

proc print data=orion.shoe_vendors n;
title 'shoe_vendors BY Group_Name Notsorted';
by Group_Name notsorted;
where Mfg_Suggested_Retail_Price>100;
var Supplier_Name Category_Name
Mfg_Suggested_Retail_Price;
run;

The NOTSORTED option turns off sequence

checking. If your data is not grouped, it can
produce a very large amount of output.

p309d11
96
 9.2 BY-Group Processing (Self-Study) 9-45

Using the NOTSORTED Option

Partial PROC PRINT Output
shoe_vendors BY Group_Name Notsorted

--------------- Group Name=Eclipse, Kid's Shoes ----------------

Supplier_ Mfg_Suggested_
Obs Name Category_Name Retail_Price

4 Eclipse Inc Children Sports $114.00

13 Eclipse Inc Children Sports $130.00
29 Eclipse Inc Children Sports $114.00
30 Eclipse Inc Children Sports $117.00

N = 4

--------------- Group Name=Tracker Kid's Clothes ---------------

Supplier_ Mfg_Suggested_
Obs Name Category_Name Retail_Price

43 3Top Sports Children Sports $106.00

55 3Top Sports Children Sports $113.00
59 3Top Sports Children Sports $109.00

N = 3

------------------------ Group Name=LSF ------------------------

Mfg_Suggested_
Obs Supplier_Name Category_Name Retail_Price

68 Eclipse Inc Clothes $141.00

97
9-46 Chapter 9 Sorting SAS Data Sets

Features of the NOTSORTED Option

The NOTSORTED option has the following features:
„ can appear anywhere in the BY statement
„ is useful if you have data that falls into other logical
groupings, such as chronological order or categories
„ can be used to access First.variable or Last.variable
in the DATA step
„ cannot be used with the MERGE and UPDATE
statements
General form of the NOTSORTED option:
BY variable-name NOTSORTED;

98

The BYSORTED SAS system option can be used to affect how SAS treats all SAS data sets.
The BYSORTED SAS system option has the following characteristics:
• specifies that observations in a data set or data sets are sorted in alphabetic or numeric order
• should be used if the data set is ordered by the BY variable

OPTIONS BYSORTED;

If observations with the same BY value are grouped together but are not necessarily sorted in alphabetic
or numeric order, use the NOBYSORTED option.

OPTIONS NOBYSORTED;

The default is BYSORTED.

 When the NOBYSORTED option is specified, you do not have to specify the NOTSORTED
option in a BY statement to access grouped data.
 9.2 BY-Group Processing (Self-Study) 9-47

9.14 Quiz
Open and submit the program p309a06.
proc print data=orion.order_fact(obs=10);
title 'Using the NOTSORTED Option with Ungrouped Data';
by Customer_ID notsorted;
var Order_ID Order_Date Delivery_Date Quantity –-
CostPrice_Per_Unit;
run;

1. Was the data grouped?

2. What do you notice about the output?

100

Business Scenario
Create a SAS data set that Partial Listing of
contains the total quantity of orion.order_fact
items sold each year. Order_Date Quantity
The values of Order_Date 11JAN2003 1
and Quantity are in the 15JAN2003 1
data set orion.order_fact. 20JAN2003 1
28JAN2003 2
27FEB2003 3
02MAR2003 1
03MAR2003 1
03MAR2003 2
09MAR2003 2
09MAR2003 1
15MAR2003 2

103
9-48 Chapter 9 Sorting SAS Data Sets

Using the GROUPFORMAT Option

data yr_totals;
keep Order_Date YrTot;
set orion.order_fact(keep=Order_Date
Quantity);
format Order_Date year4.;
by groupformat Order_Date;
if first.Order_Date then YrTot=0;
YrTot + Quantity;
if last.Order_Date;
run;

p309d12
104

The GROUPFORMAT option enables the BY statement to use the YEAR4. format to create
FIRST.Order_Date and LAST.Order_Date.

 The NOTSORTED option can be used with the GROUPFORMAT option if the data is grouped,
but not sorted.

9.15 Quiz
Open and submit the program p309a07.
data yr_totals;
keep Order_Date YrTot;
set orion.order_fact(keep=Order_Date
Quantity);
format Order_Date year4.;
by groupformat Order_Date;
if first.Order_Date then YrTot=0;
YrTot + Quantity;
if last.Order_Date;
run;
proc print data=yr_totals;
title 'Total Quantity Sold each Year';
run;

How many observations are in the data set yr_totals?

106
 9.2 BY-Group Processing (Self-Study) 9-49

The GROUPFORMAT Option

General form of the GROUPFORMAT option:
BY GROUPFORMAT variable-name <NOTSORTED>;

Advantages Disadvantages
can be used to create requires that the data set be sorted
ordered/grouped reports by the GROUPFORMAT variable or
without sorting the data grouped by the formatted values of
the GROUPFORMAT variable
causes the DATA step to available only in the DATA step
process formatted BY values
in the same way that SAS
procedures do
frequently eliminates the need
for another step
109
9-50 Chapter 9 Sorting SAS Data Sets

Using the CLASS Statement in a Procedure

Instead of using a BY statement to group data, you can
use the CLASS statement to specify the variables whose
values define the subgroup combinations for analysis by
a SAS procedure.
General form of the CLASS statement:
CLASS variable(s) </ options>;

You can use the CLASS statement with the following

Base SAS procedures:
„ PROC MEANS

„ PROC TABULATE

„ PROC SUMMARY

„ PROC UNIVARIATE
110

Reference Information

Selected options for the CLASS statement are as follows:

ORDER= specifies the order in which to group the levels of the class variables in the output.
The values for ORDER= can be any of the following:
INTERNAL orders values by ascending unformatted values. The
INTERNAL order yields the same order as the SORT procedure.
The order depends on your operating environment. This sort
sequence is particularly useful for displaying dates
chronologically. The term UNFORMATTED is an alias for
INTERNAL. INTERNAL is the default order.
DATA orders values according to their order in the input data set.
FORMATTED orders values by the ascending formatted values. This order
depends on your operating environment.
FREQ orders values by descending frequency count.

DESCENDING specifies to sort the class variable values in descending order.

GROUPINTERNAL specifies not to apply formats to the class variables when the MEANS,
SUMMARY, or TABULATE procedures group the values to create combinations
of class variables.

MISSING considers missing values as valid class variable levels. Special missing values that
represent numeric values (the letters A through Z and the underscore (_)
character) are each considered as a separate value.
 9.2 BY-Group Processing (Self-Study) 9-51

Using the CLASS Statement

What are the differences between using a BY statement
and using a CLASS statement in a procedure?
BY Statement CLASS Statement
The data set must be sorted or The data does not need to be
indexed on the BY variables. sorted or indexed on the CLASS
variables.

BY-group processing holds only The CLASS statement

one BY group in memory at a accumulates aggregates for all
time. CLASS groups simultaneously in
memory.
A percentage for the entire A percentage for the entire
report cannot be calculated with report can be calculated with
procedures such as PROC procedures such as PROC
REPORT or PROC TABULATE. REPORT or PROC TABULATE.

111

9.16 Quiz
1. Open and submit the program p309a08.
2. Change the BY statement to a CLASS statement and
resubmit the program.
3. Are the statistics created with a CLASS statement
equal to those created with a BY statement?

proc means data=orion.order_fact mean median

maxdec=2;
format Order_Date year4.;
class Order_Date;
var Quantity -- CostPrice_Per_Unit;
run;

113
9-52 Chapter 9 Sorting SAS Data Sets

Using the SUMSIZE= Memory Option

The SUMSIZE= option specifies a limit on the amount of
memory that is available for data summarization
procedures when there are class variables.

SUMSIZE=n | nK | nM | nG | nT | hexX | MIN | MAX

 The SUMSIZE= system option affects the Base SAS

procedures such as the MEANS, REPORT,
SUMMARY, and TABULATE procedures.

Proper specification of SUMSIZE= can improve

procedure performance by restricting the swapping
of memory that is controlled by the operating
environment.
115

The SUMSIZE= option is available as both a SAS system option and as a PROC statement option.
 9.2 BY-Group Processing (Self-Study) 9-53

Exercises

Level 1

4. Using the GROUPFORMAT Option

a. Open the program p309e04 that creates a format named salaryfmt.
p309e04
proc format;
value salaryfmt
.='Missing'
low - 50000='Up to $50,000'
50000 <- 100000='$50,000+ to $100,000'
100000 <- high='More than $100,000';
run;
b. Write a DATA step to do the following:
• Create a data set named payroll.
• Read only the Salary variable from the orion.employee_payroll data set.
• Group the data by Salary using the SALARYFMT format.
Hint: Sort the data by Salary before the DATA step.
• Create a variable named TotalSal that accumulates the value of Salary.
• Create a variable named Count that accumulates number of observations for each group.
• Create a variable named AvgSalary that is the average of Salary for each group.
• Format TotalSal and AvgSalary with dollar signs and two digits to the right of the decimal
point.
c. Print the data set.
Obs Salary AvgSalary TotalSal Count

1 Up to $50,000 $30,565.77 $11,309,335.00 370

2 $50,000+ to $100,000 $64,075.56 $2,883,400.00 45
3 More than $100,000 $215,207.22 $1,936,865.00 9
9-54 Chapter 9 Sorting SAS Data Sets

Level 2

5. Creating BY Groups with PROC TABULATE

a. Open the p309e05 program that contains the PROC TABULATE step below.
p309e05
options ls=80;

proc tabulate data=purchased_products format=comma12.2;

where Supplier_Name contains 'Sports';
by descending Order_Type;
class Customer_Age_Group Supplier_Name;
var Quantity Total_Retail_Price;
table Supplier_Name,
Customer_Age_Group * Total_Retail_Price=' '*sum=' '
/ printmiss misstext='$0' box='Total Retail Price';
title 'Products by Sales Supplier and Customer Age Group';
run;
b. Add a PROC SORT step to sort the data set orion.purchased_products by Order_Type and
create a temporary SAS data set named purchased_products so that the PROC TABULATE step
creates the desired output shown below.
PROC TABULATE Output
Page 1
Products by Sales Supplier and Customer Age Groups

--------------------------------- Order Type=3 ---------------------------------

„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚Total Retail Price‚ Customer Age Group ‚
‚ ‡ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚ ‚15-30 years ‚31-45 years ‚46-60 years ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Supplier Name ‚ ‚ ‚ ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰ ‚ ‚ ‚
‚3Top Sports ‚ 8,923.17‚ 8,728.44‚ 4,631.40‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Greenline Sports ‚ ‚ ‚ ‚
‚Ltd ‚ 1,232.00‚ 1,767.18‚ 1,474.08‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Pro Sportswear Inc‚ 5,684.60‚ 2,863.30‚ 2,623.10‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Top Sports ‚ $0‚ 355.20‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Triple Sportswear ‚ ‚ ‚ ‚
‚Inc ‚ $0‚ 18.20‚ 18.20‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒŒ
 9.2 BY-Group Processing (Self-Study) 9-55

Page 2
Products by Sales Supplier and Customer Age Group

--------------------------------- Order Type=2 ---------------------------------

„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚Total Retail Price‚ Customer Age Group ‚
‚ ‡ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚ ‚15-30 years ‚31-45 years ‚46-60 years ‚61-75 years ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Supplier Name ‚ ‚ ‚ ‚ ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰ ‚ ‚ ‚ ‚
‚3Top Sports ‚ 8,211.90‚ 7,512.20‚ 6,225.16‚ 4,445.20‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Greenline Sports ‚ ‚ ‚ ‚ ‚
‚Ltd ‚ 1,255.80‚ 998.30‚ 875.58‚ 1,423.90‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Pro Sportswear Inc‚ 5,797.40‚ 2,115.50‚ 2,148.30‚ 2,446.60‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Top Sports ‚ $0‚ 88.80‚ $0‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Triple Sportswear ‚ ‚ ‚ ‚ ‚
‚Inc ‚ 161.80‚ 1,001.00‚ 254.50‚ 53.80‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒŒ

c. Modify the program so that the PROC SORT step is not necessary and the PROC TABULATE
step reads from the data set orion.purchased_products. Resubmit the program.

 If necessary, consult SAS OnlineDoc or the SAS Help facility about PROC TABULATE
in order to determine what changes must be made to the program.
PROC TABULATE Output
Products by Sales Supplier and Customer Age Group

Order Type 3
„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚Total Retail Price‚ Customer Age Group ‚
‚ ‡ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚ ‚15-30 years ‚31-45 years ‚46-60 years ‚61-75 years ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Supplier Name ‚ ‚ ‚ ‚ ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰ ‚ ‚ ‚ ‚
‚3Top Sports ‚ 8,923.17‚ 8,728.44‚ 4,631.40‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Greenline Sports ‚ ‚ ‚ ‚ ‚
‚Ltd ‚ 1,232.00‚ 1,767.18‚ 1,474.08‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Pro Sportswear Inc‚ 5,684.60‚ 2,863.30‚ 2,623.10‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Top Sports ‚ $0‚ 355.20‚ $0‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Triple Sportswear ‚ ‚ ‚ ‚ ‚
‚Inc ‚ $0‚ 18.20‚ 18.20‚ $0‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒŒ
9-56 Chapter 9 Sorting SAS Data Sets

Page 2
Products by Sales Supplier and Customer Age Group

Order Type 2
„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚Total Retail Price‚ Customer Age Group ‚
‚ ‡ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚ ‚15-30 years ‚31-45 years ‚46-60 years ‚61-75 years ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Supplier Name ‚ ‚ ‚ ‚ ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰ ‚ ‚ ‚ ‚
‚3Top Sports ‚ 8,211.90‚ 7,512.20‚ 6,225.16‚ 4,445.20‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Greenline Sports ‚ ‚ ‚ ‚ ‚
‚Ltd ‚ 1,255.80‚ 998.30‚ 875.58‚ 1,423.90‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Pro Sportswear Inc‚ 5,797.40‚ 2,115.50‚ 2,148.30‚ 2,446.60‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Top Sports ‚ $0‚ 88.80‚ $0‚ $0‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚Triple Sportswear ‚ ‚ ‚ ‚ ‚
‚Inc ‚ 161.80‚ 1,001.00‚ 254.50‚ 53.80‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒŒ

Level 3

6. Sorting Data Multiple Ways

There are two additional techniques that you can use to sort data:
• using a DATA step hash object and the OUTPUT method
• using the SQL procedure ORDER BY clause
a. Starting a new SAS session between each of these tasks, benchmark the results of sorting the data
set named temp. Turn on the appropriate option to record the resources used.
The program p309e06 creates a temporary data set named temp with three variables and four
million observations.
p309e06
data temp;
do I=1 to 1000;
do J=1 to 1000;
do Name='Mary','Sue','Bob','Ted';
output;
end;
end;
end;
run;
 9.2 BY-Group Processing (Self-Study) 9-57

b. Open the program p309e06, add a PROC SORT step to create a data set named sorted_sort that
is sorted by Name, and submit the program. Record the usage statistics.
CPU
Memory
I/O

 I/O statistics are not available in the Windows operating system.

c. Open the program p309e06, add a PROC SQL step to create a new table in sorted order for Name
from temp, and submit the program. Record the usage statistics.
CPU
Memory
I/O

 I/O statistics are not available in the Windows operating system.

d. Open the program p309e06, write a DATA step using a hash object to sort the data, and submit
the program. Record the usage statistics.
CPU
Memory
I/O

 I/O statistics are not available in the Windows operating system.

9-58 Chapter 9 Sorting SAS Data Sets

9.3 Chapter Review

Chapter Review
1. Define a threaded sort.

2. What is the NOEQUALS option in the PROC SORT

statement?

3. What is the purpose of the DUPOUT= option?

118
 9.4 Solutions 9-59

9.4 Solutions

Solutions to Exercises
1. Using the PRESORTED Option
a. Use PROC CONTENTS to determine whether the data set orion.holidays is sorted.
p309s01
proc contents data=orion.holidays;
run;
b. Write a PROC SORT step to sort the data orion.holidays by Date. Create a temporary data set
named holidays. Use the PRESORTED option in the PROC SORT statement.
What is the resulting message in the log?
p309s01
proc sort data=orion.holidays out=holidays presorted;
by Date;
run;

294 proc sort data=orion.holidays out=holidays presorted;

295 by Date;
296 run;

NOTE: Sort order of input data set has been verified.

NOTE: Input data set is already sorted; it has been copied to the output data set.
NOTE: There were 364 observations read from the data set ORION.HOLIDAYS.
NOTE: The data set WORK.HOLIDAYS has 364 observations and 2 variables.
NOTE: PROCEDURE SORT used (Total process time):
real time 0.01 seconds
cpu time 0.01 seconds

c. Submit a PROC CONTENTS step to determine whether the data set holidays is sorted by Date.
p309s01
proc contents data=holidays;
run;
9-60 Chapter 9 Sorting SAS Data Sets

d. Change the BY variable to Holiday_Name and resubmit the PROC SORT step.
What is the resulting message in the log?
p309s01
proc sort data=orion.holidays out=holidays presorted;
by Holiday_Name;
run;

297 proc sort data=orion.holidays out=holidays presorted;

298 by Holiday_Name;
299 run;

NOTE: Input data set is not in sorted order.

NOTE: There were 364 observations read from the data set ORION.HOLIDAYS.
NOTE: The data set WORK.HOLIDAYS has 364 observations and 2 variables.
NOTE: PROCEDURE SORT used (Total process time):
real time 0.01 seconds
cpu time 0.01 seconds

e. Submit a PROC CONTENTS step to determine if the data set holidays is sorted by
Holiday_Name.
p309s01
proc contents data=holidays;
run;
2. Creating a Sorted Data Set
a. Open the program p309e02.
b. Modify the program to create a data set named profit07 and specify that it is sorted by Company
without sorting the data set.
p309s02
data profit07(sortedby=Company);
infile 'profit07.dat' dlm=',';
input Company : $30. Sales Cost Salaries Profit;
run;
c. Use PROC CONTENTS to verify that profit07 has a sort flag on the variable Company.
p309s02
proc contents data=profit07;
run;
 9.4 Solutions 9-61

d. Use PROC PRINT to create a report grouped by the variable Company.

p309s02
proc print data=profit07(obs=24);
by Company;
run;
e. Use the PROC SORT PRESORTED option to turn on the validated flag, and use
PROC CONTENTS to verify that the sort flag is set in the descriptor portion.
p309s02
proc sort data=profit07 presorted;
by Company;
run;

proc contents data=profit07;

run;
3. Using the SORTSEQ=LINGUISTIC Option
a. Open and submit the program p309e03.
What is the value of the LOCALE option? The answer is dependent on your site.
p309e03
proc options option=locale;
run;
b. Write a PROC SORT step to sort orion.spain_customers by Customer_Name.
c. Create an output data set named s_customers.
d. Use the SORTSEQ=LINGUISTIC option with the LOCALE=es_ES suboption.
e. Print the data set s_customers and look at the order of the variable Customer_Name.
p309s03
proc sort data=orion.spain_customers out=s_customers
sortseq=linguistic(locale=es_ES);
by Customer_Name;
run;

proc print data=s_customers;

var Customer_Name;
run;
9-62 Chapter 9 Sorting SAS Data Sets

4. Using the GROUPFORMAT Option

a. Open the program p309e04 that creates a format named salaryfmt.

b. Write a DATA step to do the following:

• Create a data set named Payroll.
• Read only the Salary variable from the orion.employee_payroll data set.
• Group the data by Salary using the salaryfmt format.
Hint: Sort the data by Salary before the DATA step.
• Create a variable named TotalSal that accumulates the value of Salary.
• Create a variable named Count that accumulates number of observations for each group.
• Create a variable named AvgSalary that is the average of Salary for each group.
• Format TotalSal and AvgSalary with dollar signs and two digits to the right of the decimal
point.

c. Print the data set.

p309s04
proc sort data=orion.employee_payroll out=employee_payroll;
by Salary;
run;

data payroll;
set employee_payroll(keep=Salary);
by Salary groupformat;
format Salary salaryfmt. AvgSalary TotalSal dollar15.2;
if first.Salary then do;
TotalSal=0;
Count=0;
end;
TotalSal+Salary;
Count+1;
if last.Salary then do;
AvgSalary=TotalSal/Count;
output;
end;
run;

proc print data=payroll;

run;
 9.4 Solutions 9-63

5. Creating BY Groups with PROC TABULATE

a. Open the p309e05 program that contains the PROC TABULATE step.
b. Add a PROC SORT step to sort the data set orion.purchased_products by Order_Type and
create a temporary SAS data set named purchased_products so that the PROC TABULATE step
creates the desired output.
p309s05
options ls=80;
proc sort data=orion.purchased_products
out=purchased_products;
by descending Order_Type;
run;

proc tabulate data=purchased_products format=comma12.2;

where Supplier_Name contains 'Sports';
by descending Order_Type;
class Customer_Age_Group Supplier_Name;
var Quantity Total_Retail_Price;
table Supplier_Name,
Customer_Age_Group * Total_Retail_Price=' '*sum=' '
/ printmiss misstext='$0' box='Total Retail Price';
title 'Products by Sales Supplier and Customer Age Group';
run;
c. Modify the program so that the PROC SORT step is not necessary and the PROC TABULATE
step reads from the data set orion.purchased_products. Resubmit the program.
p309s05
proc tabulate data=orion.purchased_products format=comma12.2;
where Supplier_Name contains 'Sports';
class Supplier_Name Customer_Age_Group ;
class Order_Type / descending;
var Quantity Total_Retail_Price;
table Order_Type,
Supplier_Name,
Customer_Age_Group * Total_Retail_Price=' '*sum=' '
/ printmiss misstext='$0' box='Total Retail Price';
title 'Products by Sales Supplier and Customer Age Group';
run;
9-64 Chapter 9 Sorting SAS Data Sets

6. Sorting Data Multiple Ways

a. Starting a new SAS session between each of these tasks, benchmark the results of sorting the data
set named temp. Turn on the appropriate option to record the resources used.
The program p309e06 creates a temporary data set named temp with three variables and four
million observations.
b. Open the program p309e06, add a PROC SORT step to create a data set named sorted_sort that
is sorted by Name, and submit the program. Record the usage statistics.
p309s06
proc sort data=temp out=sorted_sort;
by Name;
run;
CPU Answers vary according to operating environment.
Memory Answers vary according to operating environment.
I/O Answers vary according to operating environment.

 I/O statistics are not available in the Windows operating system.

c. Open the program p309e03, add a PROC SQL step to create a new table in sorted order for Name
from temp, and submit the program. Record the usage statistics.
p309s06
proc sql;
create table sorted_sql as
select *
from temp
order by Name;
quit;
CPU Answers vary according to operating environment.
Memory Answers vary according to operating environment.
I/O Answers vary according to operating environment.

 I/O statistics are not available on the Windows operating system.

 9.4 Solutions 9-65

d. Open the program p309e06, write a DATA step using a hash object to sort the data, and submit
the program. Record the usage statistics.
p309s06
data _null_;
length Name $4 I J 8;
if _N_=1 then do;
declare hash S(dataset:'temp', ordered:'Ascending');
S.definekey('Name', 'I', 'J');
S.definedata('Name', 'I', 'J');
S.definedone();
call missing(Name, I, J);
end;
S.output(dataset:'sorted_hash');
run;
CPU Answers vary according to operating environment.
Memory Answers vary according to operating environment.
I/O Answers vary according to operating environment.

 I/O statistics are not available on the Windows operating system.

9-66 Chapter 9 Sorting SAS Data Sets

Solutions to Student Activities (Polls/Quizzes)

9.01 Quiz – Correct Answer

Open and submit the program p309a01.
How many CPUs are available in your SAS session?

proc options option=cpucount;

run;

The answer varies based on the operating

environment.

14

9.04 Multiple Answer Poll – Correct Answers

If you sort a SAS data set that is in sorted order, does
SAS re-sort the data?
a. No, if the sort indicator is 'YES', and the validation
is 'YES'
b. No, if the sort indicator is 'YES', and the validation
is 'NO'
c. Yes, regardless of the sort indicator or validation
d. Yes, if the sort indicator is 'YES', and the validation
is 'NO'

31
 9.4 Solutions 9-67

9.05 Multiple Choice Poll – Correct Answer

What would be the effect of using the PRESORTED
option in the SORT procedure step in p309d05?
a. The data set January would be sorted and the
sort validation flag set to 'YES'.
b. The data set January would not be sorted and the
sort validation flag set to 'YES'.
c. The data set January would not be sorted and the
sort validation flag set to 'NO'.

42

9.06 Quiz – Correct Answer

The data set orion.salesstaff was previously sorted in
ascending order by Employee_Hire_Date.
Open and submit program p309a02.
proc sort data=orion.salesstaff
out=salesstaff;
by Employee_ID;
run;
proc print data=salesstaff;
where Employee_ID=120134;
run;

For employee number=120134, which date is first?

a. The first date hired The original order of the
data is maintained within
b. The last date hired
the new sort order because
of the default EQUALS
47
option.
9-68 Chapter 9 Sorting SAS Data Sets

9.07 Multiple Choice Poll – Correct Answer

If the data set orion.customer is sorted by
Customer_Address, which of the following values would
precede the other value in the new ordered data set?
a. 21 Hotham Parade
b. 9 Angourie Court

Because 2 is less than 9, the 21 Hotham Parade would

be the first in the sorted data set.

55

9.08 Multiple Choice Poll – Correct Answer

Which of the following would be a good use of an index
for BY-group processing?
a. Your data set is sorted by the BY variable already.
b. Your data set is updated using a SET statement in the
DATA step. In that case, the index would be updated
after the append, and you would not have to reindex
the data.
c. Your data is updated using PROC APPEND, PROC
SQL, or the MODIFY statement in the DATA step. In
that case, the index would be updated after the
update, and you would not have to reindex the data.

66
 9.4 Solutions 9-69

9.09 Multiple Choice Poll – Correct Answer

How is orion.order_fact read?
a. Direct access via the index
b. Sequentially

77

9.10 Multiple Choice Poll – Correct Answer

How is orion.street_code read?
a. Direct access via the index
b. Sequentially

79
9-70 Chapter 9 Sorting SAS Data Sets

9.11 Quiz – Correct Answer

Open and submit the program p309a03.
Are there any data errors in the log? No
data addresses;
set orion.order_fact(keep=Customer_ID Street_ID
Delivery_Date Order_ID Product_ID Quantity);
set orion.street_code(keep=Street_ID Country
Street_Name City_Name Postal_Code)
key=Street_ID / unique;
run;

86

9.12 Quiz – Correct Answer

Open and submit the program p309a04.
1. What does the SAS log show for this DATA step?
Partial SAS Log
940 data addresses2;
941 set orion.order_fact(keep=Customer_ID Street_ID Delivery_Date Order_ID Product_ID
941! Quantity) ;
942 set orion.street_code(keep=Street_ID Country Street_Name City_Name Postal_Code)
943 key=Street_ID;
944 run;

Customer_ID=23 Street_ID=9260126679 Delivery_Date=08MAR2003 Order_ID=1230338566

Product_ID=240800200063 Quantity=2 Country=US Street_Name=Ferdilah Ln Postal_Code=76092
City_Name=Tarrant _ERROR_=1 _IORC_=1230015 _N_=8
Customer_ID=45 Street_ID=9260104847 Delivery_Date=11MAR2003 Order_ID=1230371142
Product_ID=240500200003 Quantity=1 Country=US Street_Name=Angier Rd Postal_Code=94520
City_Name=Contra Costa _ERROR_=1 _IORC_=1230015 _N_=10
<lines omitted>
Customer_ID=2806 Street_ID=8010100089 Delivery_Date=01SEP2003 Order_ID=1231316727
Product_ID=240100400143 Quantity=2 Country=ZA Street_Name=Quinn Street Postal_Code=2001
City_Name=Newtown _ERROR_=1 _IORC_=1230015 _N_=82
ERROR: Limit set by ERRORS= option reached. Further errors of this type will not be printed.
Customer_ID=36 Street_ID=9260128237 Delivery_Date=17SEP2003 Order_ID=1231414059
Product_ID=240800200008 Quantity=1 Country=US Street_Name=Halstead Cir Postal_Code=55033
City_Name=Washington _ERROR_=1 _IORC_=1230015 _N_=86

88
 9.4 Solutions 9-71

9.12 Quiz – Correct Answer

Open and submit the program p309a04.
1. What does the SAS log show for this DATA step?
2. Why do you get those error messages?

Without the UNIQUE option, the following conditions

exist:
„ The value of _IORC_ ne 0.

„ The value of _ERROR_ =1.

„ The log contains data error messages.

89

9.13 Multiple Choice Poll – Correct Answer

Which of the following is true of the data set
orion.shoe_vendors?
a. The data set is sorted by the variable Group_Name.
b. The data set is not sorted by the variable
Group_Name. However, it is grouped by
Group_Name.
c. The data set is neither sorted nor grouped by the
variable Group_Name.

95
9-72 Chapter 9 Sorting SAS Data Sets

9.14 Quiz – Correct Answer

Open and submit the program p309a06.
proc print data=orion.order_fact(obs=10);
title 'Using the NOTSORTED Option with Ungrouped Data';
by Customer_ID notsorted;
var Order_ID Order_Date Delivery_Date Quantity –-
CostPrice_Per_Unit;
run;

1. Was the data grouped?

No
2. What do you notice about the output?
Some of the values of Customer_ID are repeated.
Because the NOTSORTED option turns off sequence
checking, every time that the value of Customer_ID
changes, a new BY group is created.
102

9.15 Quiz – Correct Answer

Open and submit the program p309a07.
data yr_totals;
keep Order_Date YrTot;
set orion.order_fact(keep=Order_Date
Quantity);
format Order_Date year4.;
by groupformat Order_Date;
if first.Order_Date then YrTot=0;
YrTot + Quantity;
if last.Order_Date;
run;
proc print data=yr_totals;
title 'Total Quantity Sold each Year';
run;

How many observations are in the data set yr_totals? 5

107
 9.4 Solutions 9-73

9.15 Quiz – Correct Answer

proc print data=yr_totals;
title 'Total Quantity Sold each Year';
run;

PROC PRINT Output

Total Quantity Sold each Year

Order_ Yr
Obs Date Tot

1 2003 233
2 2004 182
3 2005 153
4 2006 225
5 2007 285

108

9.16 Quiz – Correct Answer

1. Open and submit the program p309a08.
2. Change the BY statement to a CLASS statement and
resubmit the program.
3. Are the statistics created with a CLASS statement
equal to those created with a BY statement?
Yes
proc means data=orion.order_fact mean median
maxdec=2;
format Order_Date year4.;
class Order_Date;
var Quantity -- CostPrice_Per_Unit;
run;

114
9-74 Chapter 9 Sorting SAS Data Sets

Solutions to Chapter Review

Chapter Review Answers

1. Define a threaded sort.
On multiprocessor systems, the threaded sort
uses multiple processors to execute the sort
processing through simultaneous execution.
2. What is the NOEQUALS option in the PROC SORT
statement?
The NOEQUALS option does not necessarily
preserve the original order of the input data set in
the output data set.
3. What is the purpose of the DUPOUT= option?
The DUPOUT= option specifies the output data set
to which duplicate observations are written when
used in conjunction with the NODUPKEY option.
119
Chapter 10 Programmer Efficiency

10.1 Introduction................................................................................................................... 10-3

10.2 Writing Flexible Programs: Combining Raw Data Files Vertically ......................... 10-10
Exercises ............................................................................................................................ 10-46

10.3 Creating Views ............................................................................................................ 10-51

Demonstration: Creating a DATA Step View ...................................................................... 10-54

Exercises ............................................................................................................................ 10-74

10.4 Using FILE and PUT Statements to Create a SAS Program File ............................ 10-78
Demonstration: Using the DATA Step to Send E-Mail ....................................................... 10-87

Exercises ............................................................................................................................ 10-90

10.5 Using the FCMP Procedure (Self-Study) .................................................................. 10-95

Demonstration: Creating and Using Functions ................................................................ 10-106

Demonstration: Creating Subroutines Using PROC FCMP ............................................. 10-110

Exercises .......................................................................................................................... 10-115

10.6 Chapter Review......................................................................................................... 10-119

10.7 Solutions ................................................................................................................... 10-121

Solutions to Exercises ...................................................................................................... 10-121

Solutions to Student Activities (Polls/Quizzes) ................................................................. 10-134

Solutions to Chapter Review ............................................................................................ 10-140

10-2 Chapter 10 Programmer Efficiency
 10.1 Introduction 10-3

10.1 Introduction

Objectives
„ List various programming techniques for improving
programmer efficiency.
„ Provide examples of using the macro facility.
„ Use functions.
„ Substitute a procedure for a DATA step.

Techniques for Programmer Efficiency

„ Use the macro facility for repetitive coding tasks.
„ Select appropriate functions for processing data.
„ Use procedures instead of the DATA step where
possible.
„ Combine multiple steps into one step, where possible.
„ Write flexible programs.
„ Create SAS views instead of SAS files.
„ Use FILE and PUT statements to write programs
that are data-dependent.

4
10-4 Chapter 10 Programmer Efficiency

10.01 Multiple Answer Poll

Do you use any of the following features of the macro
facility?
a. Macro variables
b. Macro definitions
c. Neither of these

6
 10.1 Introduction 10-5

Using the Macro Facility

The macro facility is available to make your programs
more flexible, dynamic, and easy to maintain.
The macro facility is comprised of two components:
„ macro variables

„ macro definitions

 The macro facility is a powerful tool. However, a

detailed discussion of using macro variables and
macro definitions is beyond the scope of this
course.

Using the Macro Facility

Example of using a macro variable:

%let list=5, 9, 18, 31, 56;

proc print data=orion.customer;

where Customer_ID in (&list);
title "Customer ID Values of &list";
run;

p310d01
8
10-6 Chapter 10 Programmer Efficiency

Using the Macro Facility

Using a macro definition is a two-step process.
1. Create the macro definition.
%macro PrintSubset(dsname, var, list);

proc print data=&dsname;

where &var in (&list);
title "&var Values of &list";
run;

%mend PrintSubset;

2. Call the macro definition

%PrintSubset(orion.customer, Customer_ID,
%str(5, 9, 18, 31, 56))
p310d02
9

The %STR function masks (that is, removes the normal meaning of) these special tokens:
+ - * / , < > = ; ' "
LT EQ GT LE GE NE AND OR NOT
blank
General form of the %STR function:

%STR(argument)

argument can be any combination of text and macro triggers.

 10.1 Introduction 10-7

10.02 Quiz
In addition to saving programmer time, does creating a
macro variable or a macro definition always save
computer resources?

Why or why not?

11
10-8 Chapter 10 Programmer Efficiency

Selecting Appropriate Functions

Example of selecting appropriate functions for data
processing:
Use one of the CAT functions…
data description;
set orion.organization_dim;
Employment_Description=catx(' - ', of Company -- Job_Title);
run;

…instead of the concatenation operator and the TRIM

function.
data description;
set orion.organization_dim;
Employment_Description=trim(Company)||' - '||
trim(Department)||' - '||
trim(Section)||' - '||
trim(Org_Group)||' - '||
trim(Job_Title);
run;
p310d03
14

 The concatenation operator can be !! or ||.

The SAS®9 CAT functions are listed in the table below:

Function Use Example

CAT concatenates character strings without newvar=cat(var1,var);

removing leading or trailing blanks.

CATQ concatenates character or numeric values by newvar=catq('N','var1','var')

using a delimiter to separate items and by
adding quotation marks to strings that contain
the delimiter.

CATS concatenates character strings and removes newvar=cats(var1,var);

leading and trailing blanks.

CATT concatenates character strings and removes newvar=catt(var1,var);

trailing blanks only.

CATX concatenates character strings, removes leading newvar=catx(' ',var1,var);

and trailing blanks, and inserts separators.

If you do not specify the length of the new variable, the value of the new variable returned by any
of the CAT functions has a length of 200.

 CATQ is new in SAS 9.2.

If the receiving variable is numeric, the CAT functions remove leading and trailing blanks from numeric
arguments after they format the numeric values with the BEST. format. No note is written to the log when
the BEST. format is used.
 10.1 Introduction 10-9

Using Procedures
Example of selecting appropriate procedures for data
processing:
Use the SUMMARY procedure…

proc summary data=orion.shoe_vendors nway;

var Mfg_Suggested_Retail_Price;
class Line_Name;
output out=summary(keep=Line_Name Avg_MSP)
mean=Avg_MSP;
run;

p310d04
15 continued...

Using Procedures
…instead of the DATA step.
proc sort data=orion.shoe_vendors(keep=Line_Name
Mfg_Suggested_Retail_Price
out=shoe_vendors;
by Line_Name;
run;

data sum;
keep Line_Name Avg_MSP;
set shoe_vendors;
by Line_Name;
if first.Line_Name then do;
Tot_MSP=0;
Count=0;
end;
Tot_MSP + Mfg_Suggested_Retail_Price;
if Mfg_Suggested_Retail_Price ne . then Count+1;
if last.Line_Name then do;
Avg_MSP=Tot_MSP/Count;
output;
end;
run; p310d04
16
10-10 Chapter 10 Programmer Efficiency

10.2 Writing Flexible Programs: Combining Raw Data Files

Vertically

Objectives
„ Develop a program that is flexible.
„ Using the FILENAME statement, create a SAS data
set from multiple raw data files.
„ Using the FILEVAR= option, create a SAS data set
from multiple raw data files.

18

Flexible Programming
Programs that run in a production environment should
be as flexible as possible so that there is little, if any,
editing of the program code when the program is
submitted.
These programs are often developed using the following
steps:
1. Write an initial version of the program quickly, even
if it requires editing on subsequent submissions.
2. Add syntax, such as the DATE and TIME functions,
that can extract current information.
3. Make the program as efficient as possible.

19
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-11

Business Scenario
You need to use 12 raw data files to create a SAS data
set that contains the data for the current month and the
two previous months.
The raw data files have the same file structure and similar
names: mon1.dat, mon2.dat, and so forth. They are all
comma-separated files with these fields in the same
order: Customer_ID, Order_ID, Order_Type,
Order_Date, and Delivery_Date.
Partial Listing of mon1.dat
1 1 2 2 3 3 4 4 5 5 6
1---5----0----5----0----5----0----5----0----5----0----5----0
53,1232087464,1,13JAN2007,13JAN2007
49,1232092527,1,13JAN2007,13JAN2007
34,1232161564,1,23JAN2007,23JAN2007
2618,1232173841,3,25JAN2007,30JAN2007

20

Business Scenario
Every month you need to provide reports that contain
three months of data to Orion executives. The three
months are the current month and the previous two
months (rolling quarter).

mon8 mon9 mon10 mon11 mon12

mon8 mon9 mon10 mon11 mon12

mon8 mon9 mon10 mon11 mon12

21
10-12 Chapter 10 Programmer Efficiency

Vertical Combination Methods

Raw data can be combined vertically using several
methods:
„ concatenating files using multiple INFILE statements

„ concatenating files using a FILENAME statement

„ using the FILEVAR= option to read a list of files

„ using operating system techniques

22

Use multiple INFILE statements to solve these programming tasks:

• Reading a record from one raw data file, a record from the second raw data file, a record from the third
raw data file, and so on (similar to an interleave)
• Concatenating raw data files that have different file layouts

 Operating system techniques are outside the scope of this course.

Reading Multiple Raw Data Files

To read multiple raw data files, you can use the
FILENAME statement.

23
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-13

Using the FILENAME Statement

filename MON ('mon3.dat' 'mon2.dat' 'mon1.dat');
*filename MON ('.workshop.rawdata(mon3)'
'.workshop.rawdata(mon2)'
'.workshop.rawdata(mon1)'); * z/OS;

data quarter;
infile MON dlm=',';
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : date9.;
run;

Partial Listing of mon1.dat

1 1 2 2 3 3 4 4 5 5 6
1---5----0----5----0----5----0----5----0----5----0----5----0
53,1232087464,1,13JAN2007,13JAN2007
49,1232092527,1,13JAN2007,13JAN2007
34,1232161564,1,23JAN2007,23JAN2007
2618,1232173841,3,25JAN2007,30JAN2007

p310d05
24

In Windows and UNIX, you can use the * wildcard to specify that all 12 monthly raw data files are to be
read.
filename MON ('mon*.dat');
SAS Log
764 filename MON ('mon3.dat' 'mon2.dat' 'mon1.dat'); * PC and Unix;
765 *filename MON ('.workshop.rawdata(mon3)'
766 '.workshop.rawdata(mon2)'
767 '.workshop.rawdata(mon1)'); * z/OS;
768
769 data quarter;
770 infile MON dlm=',';
771 input Customer_ID Order_ID Order_Type
772 Order_Date : date9. Delivery_Date : Date9.;
773 run;

NOTE: The infile MON is:

File Name='S:\Workshop\mon3.dat,

File List=('S:\Workshop\mon3.dat' 'S:\Workshop\mon2.dat' 'S:\Workshop\mon1.dat'),

RECFM=V,LRECL=256

NOTE: The infile MON is:

File Name= S:\Workshop\mon2.dat,

File List=('S:\Workshop\mon3.dat' 'S:\Workshop\mon2.dat' 'S:\Workshop\mon1.dat'),

RECFM=V,LRECL=256

NOTE: The infile MON is:

(Continued on the next page.)

10-14 Chapter 10 Programmer Efficiency

File Name=S:\Workshop\mon1.dat,

File List=('S:\Workshop\mon3.dat' 'S:\Workshop\mon2.dat' 'S:\Workshop\mon1.dat'),

RECFM=V,LRECL=256

NOTE: 9 records were read from the infile MON.

The minimum record length was 34.
The maximum record length was 38.
NOTE: 7 records were read from the infile MON.
The minimum record length was 35.
The maximum record length was 38.
NOTE: 4 records were read from the infile MON.
The minimum record length was 35.
The maximum record length was 37.
NOTE: The data set WORK.QUARTER has 20 observations and 5 variables.
NOTE: DATA statement used (Total process time):
real time 2.71 seconds
cpu time 0.03 seconds

774
775 proc print data=quarter;
776 title 'quarter ';
777 run;

NOTE: There were 20 observations read from the data set WORK.QUARTER.
NOTE: PROCEDURE PRINT used (Total process time):
real time 1.34 seconds
cpu time 0.00 seconds

FILENAME Statement Syntax

General form of the FILENAME statement:

FILENAME fileref ('external-file1'

'external-file2' …
'external-filen');

fileref any valid SAS name that is eight

characters or fewer
'external-file' the physical name of an external file; the
physical name is the name that is
recognized by the operating environment

25

A FILENAME statement can associate a fileref with multiple physical external files.
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-15

10.03 Quiz
1. Open and submit the program p310a01.
2. How many observations are in the data set quarter?

3. Change the files referenced by MON to the current

month and the two previous months. How many
observations are in the data set quarter?

27

INFILE Statement with FILEVAR= Option

To make the program more flexible, use the FILEVAR=
option in the INFILE statement to provide the name of the
raw data file instead of using the FILENAME statement,
which would need to be edited every month.
infile ORD filevar=NextFile;

ORD an arbitrarily named placeholder, not an

actual filename or a fileref that was assigned
to a file previously. SAS uses this
placeholder for reporting processing
information to the SAS log.
NextFile a variable whose value contains the name of
the raw data file to be read (mon9.dat,
mon10.dat, mon11.dat, and so on).
29
10-16 Chapter 10 Programmer Efficiency

INFILE Statement with FILEVAR= Option

„ The FILEVAR= option names a variable whose
change in value causes the INFILE statement to close
the current input file and open a new one.
„ When the next INPUT statement executes, it reads
from the new file that the FILEVAR= variable specifies.

General form of the FILEVAR= variable option:

INFILE file-specification FILEVAR=variable;

variable a temporary character variable that contains

the physical filename of the raw data file to
be read

30

 Similar to automatic variables, the FILEVAR= variable is not written to the data set.
The FILEVAR= variable can read raw data files conditionally.

Creating the Filename

The names of the raw data files can be constructed
programmatically.
In this example, the raw data files are named in a
consistent form.

mon + 9 + .dat
mon + 10 + .dat
mon + 11 + .dat
There are multiple techniques for creating the names of
the raw data files.

31
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-17

10.04 Quiz
If the value of the variable i is the number of the month,
which of the following could be used to create the name
of the raw data file?
a. NextFile=cats("mon",i,".dat");

b. NextFile="mon"||put(i,2.)||".dat";
NextFile=compress(NextFile);

c. NextFile=compress("mon"||put(i,2.)||".dat");

33

Creating the Filename

After the value of NextFile is created, the FILEVAR=
option identifies the raw data file to be read.
do i=11,10,9;
NextFile=cats("mon",i,".dat");
infile ORD filevar=NextFile dlm=',';
end;

When i=11
NextFile=mon11.dat
When i=10
NextFile=mon10.dat
When i=9
NextFile=mon9.dat
35
10-18 Chapter 10 Programmer Efficiency

Reading Raw Data

data movingq;
drop i;
do i=11, 10, 9;
c NextFile=cats("mon",i,".dat");
d infile ORD filevar=NextFile dlm=',';
e input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
f output;
end;
g stop;
run;
p310d06
36

The first four of the following statements are within the DO loop:
c The assignment statement creates the name of the raw data file.
d The INFILE statement with the FILEVAR= option names the raw data file. In addition, the
FILEVAR= option closes the current file and opens a new file if the value of the FILEVAR= variable
changes.
e The INPUT statement copies a record of the raw data file, converts it to SAS format, and writes it to
the PDV.
f The OUTPUT statement outputs the observation that is created by the INPUT statement.
g The STOP statement outside the DO loop stops the DATA step after all of the observations are
written.
In this example, the DATA step does not encounter the end of file. If the STOP statement were not
included, the program would continue to execute the DO loop repetitively. Therefore, the STOP statement
is needed to prevent an infinite loop of the DATA step.
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-19

SAS Log
807
808 data movingq;
809 drop i;
810 do i=11,10,9;
811 NextFile=cats("mon",put(i,2.),".dat"); * PC and UNIX;
812 *NextFile=cats(".lwprg3.rawdata(mon",put(i,2.),")"); * mainframe ;
813 infile ORD filevar=NextFile dlm=',';
814 input Customer_ID Order_ID Order_Type
815 Order_Date:date9. Delivery_Date:Date9.;
816 output;
817 end;
818 stop;
819 run;

NOTE: The infile ORD is:

File Name=S:\Workshop\mon11.dat,
RECFM=V,LRECL=256

NOTE: The infile ORD is:

File Name=S:\Workshop\mon10.dat,
RECFM=V,LRECL=256

NOTE: The infile ORD is:

File Name=S:\Workshop\mon9.dat,
RECFM=V,LRECL=256

NOTE: 1 record was read from the infile ORD.

The minimum record length was 35.
The maximum record length was 35.
NOTE: 1 record was read from the infile ORD.
The minimum record length was 36.
The maximum record length was 36.
NOTE: 1 record was read from the infile ORD.
The minimum record length was 34.
The maximum record length was 34.
NOTE: The data set WORK.MOVINGQ has 3 observations and 5 variables.
NOTE: DATA statement used (Total process time):
real time 0.18 seconds
cpu time 0.01 seconds

820
821 proc print data=movingq;
822 title 'Moving Quarter Data';
823 run;

NOTE: There were 3 observations read from the data set WORK.MOVINGQ.
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds
10-20 Chapter 10 Programmer Efficiency

Setup for the Poll

Refer to the following program for the next two polls:

data movingq;
drop i;
do i=11, 10, 9;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=',';
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : date9.;
output;
end;
stop;
run;

38

10.05 Poll
Is the STOP statement necessary?
€ Yes
€ No

39
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-21

10.06 Multiple Choice Poll

How many observations are in movingq?
a. One observation per record in all of the raw data files
b. 0
c. 1
d. 3

41

Reading Raw Data

data movingq;
drop i;
do i=11, 10, 9;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
output;
end;
end;
stop;
run;
43 p310d07

The DO WHILE statement continues to execute the INFILE statement for every record of the raw data
file until the value of LastObs=1. The DO WHILE statement checks the condition at the top of the loop.
The END= option creates the variable LastObs that can be used to determine the end of the raw data file.
The END= option names a variable whose value is one of the following:

0 when the current input data record is not the last in the current input file

1 when the current input record is the last in the current input file
10-22 Chapter 10 Programmer Efficiency

Partial SAS Log

127 data movingq;
128 drop i;
129 do i=11, 10, 9;
130 NextFile=cats("mon", i, ".dat");
131 infile ORD filevar=NextFile dlm=','
132 end=LastObs;
133 do while (not LastObs);
134 input Customer_ID
135 Order_ID
136 Order_Type
137 Order_Date : date9.
138 Delivery_Date : date9.;
139 output;
140 end;
141 end;
142 stop;
143 run;

NOTE: The infile ORD is:

Filename=S:\workshop\mon11.dat,
RECFM=V,LRECL=256,File Size (bytes)=297,
Last Modified=27Jan2008:20:33:40,
Create Time=27Jan2008:20:33:40

NOTE: The infile ORD is:

Filename= S:\workshop\mon10.dat,
RECFM=V,LRECL=256,File Size (bytes)=265,
Last Modified=27Jan2008:20:33:28,
Create Time=27Jan2008:20:33:28

NOTE: The infile ORD is:

Filename= S:\workshop\mon9.dat,
RECFM=V,LRECL=256,File Size (bytes)=109,
Last Modified=27Jan2008:20:33:16,
Create Time=27Jan2008:20:33:16

NOTE: 8 records were read from the infile ORD.

The minimum record length was 35.
The maximum record length was 36.
NOTE: 7 records were read from the infile ORD.
The minimum record length was 35.
The maximum record length was 38.
NOTE: 3 records were read from the infile ORD.
The minimum record length was 34.
The maximum record length was 35.
NOTE: The data set WORK.MOVINGQ has 18 observations and 5 variables.
NOTE: DATA statement used (Total process time):
real time 0.01 seconds
cpu time 0.01 seconds
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-23

Reading the Current Month

data movingq;
drop MonNum MidMon LastMon i;
MonNum=month(today());
MidMon=MonNum - 1;
LastMon=MidMon - 1;
do i=MonNum, Midmon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
output;
end;
end;
stop;
run;
p310d08
44

The MONTH function is used to obtain the month number of today’s date to begin the rolling month
range. The month numbers of the two months before today’s month number are then calculated.
10-24 Chapter 10 Programmer Efficiency

Setup for the Poll

The following program is submitted:
data movingq;
drop MonNum MidMon LastMon i;
MonNum=month(today());
MidMon=MonNum - 1;
LastMon=MidMon - 1;
do i=MonNum, Midmon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
output;
end;
end;
stop;
run;
p310d08
46

10.07 Poll
Will the SAS code in p310d08 produce the correct results
if the current month is January or February?
€ Yes
€ No

47
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-25

INTNX Function
The INTNX function increments a date value by a given
interval or intervals, and returns a date value.
EDate=intnx('interval', BDate, increment)
Formatted Value Using the INTNX Function Formatted Value
of BDate of EDate
04JUL2008 intnx('year', BDate, -1) 01JAN2007
04JUL2008 intnx('year', BDate, 0) 01JAN2008
04JUL2008 intnx('year', BDate, 1) 01JAN2009
04JUL2008 intnx('month', BDate, -2) 01MAY2008
04JUL2008 intnx('month', BDate, -1) 01JUN2008
04JUL2008 intnx('month', BDate, 0) 01JUL2008
04JUL2008 intnx('month', BDate, 1) 01AUG2008
04JUL2008 intnx('month', BDate, 2) 01SEP2008
49

 The INTNX function also supports multiples of an interval and shifted intervals.

The program p310d08a contains the SAS DATA step code to replicate these results.
data dates;
BDate='04JUL2008'd;
PreviousYear=intnx('year', BDate, -1);
ThisYear=intnx('year', BDate, 0);
NextYear=intnx('year', BDate, 1);
TwoMonthsBack=intnx('month', BDate, -2);
PreviousMonth=intnx('month', BDate, -1);
ThisMonth=intnx('month', BDate, 0);
NextMonth=intnx('month', BDate, 1);
TwoMonthsFromNow=intnx('month', BDate, 2);
format BDate PreviousYear ThisYear NextYear TwoMonthsBack
PreviousMonth ThisMonth NextMonth TwoMonthsFromNow date9.;
run;

proc print data=dates;

title 'INTNX Function Results';
run;
10-26 Chapter 10 Programmer Efficiency

INTNX Function
General form of the INTNX function:
INTNX('interval', start-from, increment<, alignment>)

'interval' specifies a character constant or a variable

containing a date, datetime, or time interval.
start-from specifies a SAS expression that represents
a SAS date, datetime, or time value
identifying a starting point.
increment specifies a negative or positive integer that
represents the specific number of intervals.
alignment controls the position of SAS dates within the
interval.
50

The INTNX function also supports multiples of an interval and shifted intervals.
General form of the INTNX function with multiples and shift indexes:

INTNX(interval<multiple><.shift-index>, start-from, increment <,alignment>)

interval specifies a character constant, a variable, or an expression that contains a datetime

interval such as WEEK, SEMIYEAR, QTR, or HOUR. The type of interval (date,
datetime, or time) must match the type of the value in start-from and increment.

multiple specifies a character constant, a variable, or an expression that contains a datetime

interval such as WEEK, SEMIYEAR, QTR, or HOUR. The type of interval (date,
datetime, or time) must match the type of the value in start-from and increment.

shift- specifies the starting point of the interval. By default, the starting point is 1. A value
index that is greater than 1 shifts the start to a later point within the interval. The unit for
shifting depends on the interval. For example, YEAR.3 specifies yearly periods that
are shifted to start on the first of March of each calendar year and to end in February
of the following year. The shift index cannot be greater than the number of periods in
the entire interval. For example, YEAR2.24 has a valid shift index, but YEAR2.25 is
invalid because there is no twenty-fifth month in a two-year interval. If the default
shift period is the same as the interval type, then you can shift only multi-period
intervals with the shift index. For example, because MONTH type intervals shift by
MONTH sub-periods by default, you cannot shift monthly intervals with the shift
index. However, you can shift bimonthly intervals with the shift index, because two
MONTH intervals exist in each MONTH2 interval. The interval name MONTH2.2,
for example, specifies bimonthly periods starting on the first day of even-numbered
months.
(Continued on the next page.)
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-27

start-from specifies a SAS expression that represents a SAS date, time, or datetime value that
identifies a starting point.

increment specifies a negative, positive, or zero integer that represents the number of date, time,
or datetime intervals. Increment is the number of intervals to shift the value of start-
from.

alignment controls the position of SAS dates within the interval. Alignment can be one of these
values:
The values of alignment can be as follows:

BEGINNING | B specifies that the returned date is aligned to the beginning of the
interval (DEFAULT).

MIDDLE | M specifies that the returned date is aligned to the midpoint of the
interval.

END | E specifies that the returned date is aligned to the end of the interval.

SAMEDAY | S | SAME specifies that the date that is returned is aligned to the same calendar
date with the corresponding interval increment.

 Alignment is new in SAS®9.

10-28 Chapter 10 Programmer Efficiency

Using the INTNX Function

data movingq;
drop MonNum MidMon LastMon i;
MonNum=month(today());
MidMon=month(intnx('month', today(), -1));
LastMon=month(intnx('month', today(), -2));
do i=MonNum, Midmon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
output;
end;
end;
stop;
run;
51 p310d09

 For z/OS (OS/390 ):

NextFile='.prog3.rawdata(mon'||put(i,2.)||')';
SAS Log
144 data movingq;
145 drop MonNum MidMon LastMon i;
146 MonNum=month(today());
147 MidMon=month(intnx('month', today(), -1));
148 LastMon=month(intnx('month', today(), -2));
149 do i=MonNum, Midmon, LastMon;
150 NextFile=cats("mon", i, ".dat");
151 infile ORD filevar=NextFile dlm=','
152 end=LastObs;
153 do while (not LastObs);
154 input Customer_ID
155 Order_ID
156 Order_Type
157 Order_Date : date9.
158 Delivery_Date : date9.;
159 output;
160 end;
161 end;
162 stop;
163 run;

NOTE: The infile ORD is:

Filename=S:\workshop\mon5.dat,
RECFM=V,LRECL=256,File Size (bytes)=411,
Last Modified=27Jan2008:20:32:30,
Create Time=27Jan2008:20:32:30

(Continued on the next page.)

 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-29

NOTE: The infile ORD is:

Filename= S:\workshop\mon4.dat,
RECFM=V,LRECL=256,File Size (bytes)=522,
Last Modified=27Jan2008:20:32:18,
Create Time=27Jan2008:20:32:18

NOTE: The infile ORD is:

Filename= S:\workshop\mon3.dat,
RECFM=V,LRECL=256,File Size (bytes)=334,
Last Modified=27Jan2008:20:32:04,
Create Time=27Jan2008:20:32:04

NOTE: 11 records were read from the infile ORD.

The minimum record length was 34.
The maximum record length was 36.
NOTE: 14 records were read from the infile ORD.
The minimum record length was 34.
The maximum record length was 38.
NOTE: 9 records were read from the infile ORD.
The minimum record length was 34.
The maximum record length was 38.
NOTE: The data set WORK.MOVINGQ has 34 observations and 5 variables.
NOTE: DATA statement used (Total process time):
real time 0.03 seconds
cpu time 0.03 seconds

 This program was run in May.

10.08 Quiz
p310d09 contains the following code.
MonNum=month(today());
MidMon=month(intnx('month', today(), -1));
LastMon=month(intnx('month', today(), -2));
Why is the following program more efficient?
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));

53
10-30 Chapter 10 Programmer Efficiency

Using the INTNX Function

data movingq;
drop Today MonNum MidMon LastMon i;
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));
do i=MonNum, MidMon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=',' end=LastObs;
do while (not LastObs);
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : date9.;
output;
end;
end;
stop;
run;

55 p310d10

How Does FILEVAR= Work? (Self-Study)

Suppose you had three simple raw data files.
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9

56
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-31

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs;
do while(not LastObs);
input Num1 Num2;
output;
end;
end;
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
0 . . 0 1
57 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one; i is initialized to 'a'
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs;
do while(not LastObs);
input Num1 Num2;
output;
end;
end;
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a 0 . . 0 1
58 ...
10-32 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs;
do while(not LastObs);
input Num1 Num2;
output;
end;
end;
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 0 . . 0 1
59 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs;
do while(not LastObs);
input Num1 Num2;
output;
end;
end;
stop; LastObs=0
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 0 . . 0 1
60 ...

 LastObs is reset to 0 because the value of FILEVAR= changed and a new file is opened.
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-33

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs;
do while(not LastObs);
input Num1 Num2;
output;
end; The DO WHILE evaluates
end; the condition at the top
stop; of the loop.
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 0 . . 0 1
61 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 4
do while(not LastObs);
input Num1 Num2;
output;
end;
end;
stop; LastObs=0
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 0 1 4 0 1
62 ...
10-34 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
Output current observation.
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 4
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end;
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 0 1 4 0 1
63 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 4
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop
stop; executes.
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 0 1 4 0 1
64 ...
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-35

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end;
stop; LastObs=1
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 1 1 5 0 1
65 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
Output current observation.
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 1 1 5 0 1
66 ...
10-36 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop 1 5
stop; does not execute.
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
a a.dat 1 1 5 0 1
67 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one; i increments to 'b'.
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b a.dat 1 1 5 0 1
68 ...
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-37

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b b.dat 1 1 5 0 1
69 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; LastObs=0 1 5
stop;
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b b.dat 0 1 5 0 1
70 ...
10-38 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 1 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop 1 5
stop; executes.
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b b.dat 0 1 5 0 1
71 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; LastObs=1
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b b.dat 1 2 5 0 1
72 ...
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-39

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
Output current observation.
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b b.dat 1 2 5 0 1
73 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop 1 5
stop; 2 5
run;
does not execute.

PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
b b.dat 1 2 5 0 1
74 ...
10-40 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one; i increments to 'c'.
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c b.dat 1 2 5 0 1
75 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 1 2 5 0 1
76 ...
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-41

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; LastObs=0 2 5
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 0 2 5 0 1
77 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 2 , 5
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop 1 5
stop; 2 5
run;
executes.

PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 0 2 5 0 1
78 ...
10-42 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 8
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; LastObs=0 2 5
run;
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 0 3 8 0 1
79 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
Output current observation.
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 8
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run; 3 8

PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 0 3 8 0 1
80 ...
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-43

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 8
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop 1 5
stop; 2 5
run;
executes. 3 8

PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 0 3 8 0 1
81 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 9
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; LastObs=1 1 5
stop; 2 5
run; 3 8

PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 1 3 9 0 1
82 ...
10-44 Chapter 10 Programmer Efficiency

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
Output current observation.
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 9
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run; 3 8
3 9
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 1 3 9 0 1
83 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 9
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; The DO WHILE loop 1 5
stop; 2 5
run;
does not execute. 3 8
3 9
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 1 3 9 0 1
84 ...
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-45

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c'; The values of i
NextFile=cats(i,".dat");
Input Buffer
are all assigned.
infile ORD filevar=NextFile dlm=',' 1 2 3 4 5 6
end=LastObs; 3 , 9
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run; 3 8
3 9
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 1 3 9 0 1
85 ...

Execution (Self-Study)
a b c
1 1 1
1---5----0 1---5----0 1---5----0
1, 4 2, 5 3, 8
1, 5 3, 9
data one;
do i='a', 'b', 'c';
NextFile=cats(i,".dat");
Input Buffer
The DATA step 1 2 3 4 5 6
infile ORD filevar=NextFile dlm=','
stops execution.
end=LastObs; 3 , 9
do while(not LastObs);
input Num1 Num2; one
output; Num1 Num2
end; 1 4
end; 1 5
stop; 2 5
run; 3 8
3 9
PDV
i NextFile DLastObs
D Num1 Num2 D _ERROR_ D _N_
c c.dat 1 3 9 0 1
86 ...
10-46 Chapter 10 Programmer Efficiency

Exercises

Level 1

1. Using the FILENAME Statement

The raw data files level_1, level_2, and level_3 contain customer information for the three types of
customers: Orion Club Gold members, Orion Club members, and Internet/Catalog customers. The
fields in each file are delimited with commas.
The raw data files use the naming convention level_#. For example:
• For directory based: level_1.dat
• For z/OS (OS/390): '.prog3.rawdata(level_1)'
a. Open the program p310e01, which contains the following statements:
p310e01
data all_levels;
length Customer_Name $ 40 Customer_Age_Group $ 12
Customer_Type $ 40 Customer_Group $ 40;
input Customer_Name $ Customer_Age_Group $ Customer_Type $
Customer_Group $;
run;

proc print data=all_levels;

run;
b. Use the FILENAME statement to concatenate the three raw data files.
c. Modify the DATA step to use the fileref created in part b to create the SAS data set all_levels.
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-47

d. Print the all_levels data set.

Partial PROC PRINT Output
Customer_
Obs Customer_Name Age_Group Customer_Type

1 Sandrina Stephano 15-30 years Orion Club Gold members medium activity
2 Cornelia Krahl 31-45 years Orion Club Gold members medium activity
3 Markus Sepke 15-30 years Orion Club Gold members low activity
4 Oliver S. Füßling 31-45 years Orion Club Gold members high activity
5 Cynthia Martinez 46-60 years Orion Club Gold members medium activity

Obs Customer_Group

1 Orion Club Gold members

2 Orion Club Gold members
3 Orion Club Gold members
4 Orion Club Gold members
5 Orion Club Gold members

< lines removed >

Customer_
Obs Customer_Name Age_Group Customer_Type

67 Soberina Berent 15-30 years Orion Club members medium activity

68 Alex Santinello 15-30 years Orion Club members medium activity
69 Kenan Talarr 31-45 years Orion Club members high activity
70 Ulrich Heyde 61-75 years Internet/Catalog Customers
71 Tulio Devereaux 46-60 years Internet/Catalog Customers
72 Robyn Klem 46-60 years Internet/Catalog Customers
73 Cynthia Mccluney 31-45 years Internet/Catalog Customers
74 Candy Kinsey 61-75 years Internet/Catalog Customers
75 Phenix Hill 31-45 years Internet/Catalog Customers
76 Avinoam Zweig 46-60 years Internet/Catalog Customers
77 Lauren Marx 31-45 years Internet/Catalog Customers

Obs Customer_Group

67 Orion Club members

68 Orion Club members
69 Orion Club members
70 Internet/Catalog Customers
71 Internet/Catalog Customers
72 Internet/Catalog Customers
73 Internet/Catalog Customers
74 Internet/Catalog Customers
75 Internet/Catalog Customers
76 Internet/Catalog Customers
77 Internet/Catalog Customers
10-48 Chapter 10 Programmer Efficiency

Level 2

2. Using the FILEVAR= Option to Read from Raw Data

The raw data files level_1, level_2, and level_3 contain customer information for the three types of
customers: Orion Club Gold members, Orion Club members, and Internet/Catalog customers.
The raw data files use the naming convention level_#. For example:
• For directory based: level_1.dat
• For z/OS (OS/390): '.prog3.rawdata(level_1)'
a. Open the program p310e02, which contains the following statements:
p310e02
data all_levels;
length Customer_Name $ 40 Customer_Age_Group $ 12
Customer_Type $ 40 Customer_Group $ 40;

input Customer_Name $ Customer_Age_Group $ Customer_Type $

Customer_Group $;
run;

proc print data=all_levels;

run;
b. Use the FILEVAR= option to concatenate the three raw data files and create the SAS data set
all_levels.
 10.2 Writing Flexible Programs: Combining Raw Data Files Vertically 10-49

c. Print the all_levels data set.

Partial Listing of all_levels
Customer_
Obs Customer_Name Age_Group Customer_Type

1 Sandrina Stephano 15-30 years Orion Club Gold members medium activity
2 Cornelia Krahl 31-45 years Orion Club Gold members medium activity
3 Markus Sepke 15-30 years Orion Club Gold members low activity
4 Oliver S. Füßling 31-45 years Orion Club Gold members high activity
5 Cynthia Martinez 46-60 years Orion Club Gold members medium activity

Obs Customer_Group

1 Orion Club Gold members

2 Orion Club Gold members
3 Orion Club Gold members
4 Orion Club Gold members
5 Orion Club Gold members

< lines removed >

Customer_
Obs Customer_Name Age_Group Customer_Type

67 Soberina Berent 15-30 years Orion Club members medium activity

68 Alex Santinello 15-30 years Orion Club members medium activity
69 Kenan Talarr 31-45 years Orion Club members high activity
70 Ulrich Heyde 61-75 years Internet/Catalog Customers
71 Tulio Devereaux 46-60 years Internet/Catalog Customers
72 Robyn Klem 46-60 years Internet/Catalog Customers
73 Cynthia Mccluney 31-45 years Internet/Catalog Customers
74 Candy Kinsey 61-75 years Internet/Catalog Customers
75 Phenix Hill 31-45 years Internet/Catalog Customers
76 Avinoam Zweig 46-60 years Internet/Catalog Customers
77 Lauren Marx 31-45 years Internet/Catalog Customers

Obs Customer_Group

67 Orion Club members

68 Orion Club members
69 Orion Club members
70 Internet/Catalog Customers
71 Internet/Catalog Customers
72 Internet/Catalog Customers
73 Internet/Catalog Customers
74 Internet/Catalog Customers
75 Internet/Catalog Customers
76 Internet/Catalog Customers
77 Internet/Catalog Customers
10-50 Chapter 10 Programmer Efficiency

Level 3

3. Using the FILEVAR= Option to Read Filenames from a SAS Data Set
The SAS data set orion.month_file contains the names of the raw data files that need to be
concatenated.
Listing of orion.month_file
Obs File_Name

1 mon1.dat
2 mon2.dat
3 mon3.dat
4 mon4.dat
5 mon5.dat
6 mon6.dat
7 mon7.dat
8 mon8.dat
9 mon9.dat
10 mon10.dat
11 mon11.dat
12 mon12.dat

The starter file p310e03 contains the following DATA step program:
p310e03
data all_months;
format Order_Date Delivery_Date date9.;
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : Date9.;
run;

proc print data=all_months;

run;
a. Use the FILEVAR= option to create a SAS data set named All_Months from the raw data files
named in orion.month_file.
b. Print the first 10 observations of the All_Months SAS data set.
Partial Listing of All_Months
Order_ Delivery_ Customer_ Order_
Obs Date Date ID Order_ID Type

1 13JAN2007 13JAN2007 53 1232087464 1

2 13JAN2007 13JAN2007 49 1232092527 1
3 23JAN2007 23JAN2007 34 1232161564 1
4 25JAN2007 30JAN2007 2618 1232173841 3
5 01FEB2007 04FEB2007 89 1232217725 2
6 05FEB2007 05FEB2007 195 1232240447 1
7 05FEB2007 06FEB2007 70046 1232241009 3
8 15FEB2007 15FEB2007 171 1232307056 1
9 15FEB2007 15FEB2007 20 1232311932 1
10 18FEB2007 22FEB2007 23 1232331499 2
 10.3 Creating Views 10-51

10.3 Creating Views

Objectives
„ List the types of SAS data sets.
„ Create and use DATA step views.
„ List the advantages of DATA step views.
„ List guidelines for using DATA step views.

90
10-52 Chapter 10 Programmer Efficiency

SAS Data Sets

Instead of creating a SAS data file that contains three
months of raw data, as discussed previously, you can
create a DATA step view.

SAS Data File SAS Data View

data stored instructions
on disk stored on disk

 SAS views are physically smaller than the

corresponding file would be.
91

A SAS data file… A DATA step view…

is a SAS file with a member type of DATA. is a SAS file with a member type of VIEW.

enables read or write capabilities. is read-only.

contains data and a descriptor portion that are contains no data.

stored on disk.

contains a partially compiled DATA step.

A SAS Data File

filename fileref 'ext-file';
data orion.newdata;
infile fileref; External
<additional SAS statements> File
run;

proc print data=orion.newdata;

run;

92 ...
 10.3 Creating Views 10-53

A DATA Step View

data orion.newview/
view=orion.newview;
infile fileref; External
<additional SAS statements> File
run;

Compilation Execution

filename fileref 'ext-file';

proc print data=orion.newview;
run;

93

The name of a DATA view must be different from the name of any existing SAS data file or view in the
same SAS library.
10-54 Chapter 10 Programmer Efficiency

Creating a DATA Step View

p310d11, p310d12
p310d11
data orion.quarter / view=orion.quarter;
infile MON dlm=',';
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : date9.;
run;
SAS Log
1005
1006 data orion.quarter / view=orion.quarter;
1007 infile MON dlm=',';
1008 input Customer_ID Order_ID Order_Type
1009 Order_Date : date9. Delivery_Date : Date9.;
1010 run;

NOTE: DATA STEP view saved on file ORION.QUARTER.

NOTE: A stored DATA STEP view cannot run under a different operating system.
NOTE: DATA statement used (Total process time):
real time 0.17 seconds
cpu time 0.00 seconds

p310d11
filename MON ('mon3.dat' 'mon2.dat' 'mon1.dat'); * PC and UNIX;
*filename MON ('.workshop.rawdata(mon3)'
'.workshop.rawdata(mon2)'
'.workshop.rawdata(mon1)'); * z/OS;

proc print data=orion.quarter;

title 'quarter';
run;
 10.3 Creating Views 10-55

Partial PROC PRINT Output

quarter

Customer_ Order_ Order_ Delivery_

Obs ID Order_ID Type Date Date

1 53 1232087464 1 17179 17179

2 49 1232092527 1 17179 17179
3 34 1232161564 1 17189 17189
4 2618 1232173841 3 17191 17196
5 89 1232217725 2 17198 17201
6 195 1232240447 1 17202 17202
7 70046 1232241009 3 17202 17203
8 171 1232307056 1 17212 17212
9 20 1232311932 1 17212 17212
10 23 1232331499 2 17215 17219
11 13 1232373481 2 17222 17225
12 4 1232410925 3 17227 17228
13 4 1232455720 1 17234 17234
14 19 1232478868 1 17238 17238
15 70201 1232517885 3 17244 17249
16 4 1232530384 3 17245 17246
17 49 1232530393 1 17245 17245
18 92 1232554759 1 17249 17249
19 195 1232590052 1 17255 17255
20 89 1232601472 2 17256 17259

SAS Log
1011 filename MON ('mon3.dat' 'mon2.dat' 'mon1.dat'); * PC and Unix;
1012 *filename MON ('.workshop.rawdata(mon3)'
1013 '.workshop.rawdata(mon2)'
1014 '.workshop.rawdata(mon1)'); * z/OS;
1015
1016 proc print data=orion.quarter;
1017 title ' quarter';
1018 run;

NOTE: The infile MON is:

File Name=S:\Workshop\mon3.dat,

File List=('S:\Workshop\mon3.dat' 'S:\Workshop\mon2.dat' S:\Workshop\mon1.dat'),

RECFM=V,LRECL=256

NOTE: The infile MON is:

File Name= S:\Workshop\mon2.dat,

File List=('S:\Workshop\mon3.dat' 'S:\Workshop\mon2.dat' 'S:\Workshop\mon1.dat'),

RECFM=V,LRECL=256

(Continued on the next page.)

10-56 Chapter 10 Programmer Efficiency

NOTE: The infile MON is:

File Name=S:\Workshop\mon1.dat,

File List=('S:\Workshop\mon3.dat' 'S:\Workshop\mon2.dat' 'S:\Workshop\mon1.dat'),

RECFM=V,LRECL=256

NOTE: 9 records were read from the infile MON.

The minimum record length was 34.
The maximum record length was 38.
NOTE: 7 records were read from the infile MON.
The minimum record length was 35.
The maximum record length was 38.
NOTE: 4 records were read from the infile MON.
The minimum record length was 35.
The maximum record length was 37.

NOTE: View ORION.QUARTER.VIEW used (Total process time):

real time 0.25 seconds
cpu time 0.00 seconds

NOTE: There were 20 observations read from the data set ORION.QUARTER_MON.
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.81 seconds
cpu time 0.00 seconds

p310d12
data orion.movingq / view=orion.movingq;
drop Today MonNum MidMon LastMon i;
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));
do i=MonNum, MidMon, LastMon;
NextFile=cats("mon", i, ".dat"); * Windows and UNIX;
*NextFile=cats(".workshop.rawdata(mon", i, ")"); * z/OS ;
infile ORD filevar=NextFile dlm=',' end=LastObs;
do while (not LastObs);
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : date9.;
output;
end;
end;
stop;
run;
(Continued on the next page.)
 10.3 Creating Views 10-57

SAS Log
64 data orion.movingq / view=orion.movingq;
65 drop MonNum MidMon LastMon i;
66 Today=today();
67 MonNum=month(Today);
68 MidMon=month(intnx('month', Today, -1));
69 LastMon=month(intnx('month', Today, -2));
70 do i=MonNum, MidMon, LastMon;
71 NextFile=cats("mon", i, ".dat"); * Windows and UNIX;
72 *NextFile=cats(".workshop.rawdata(mon", i, ")"); * z/OS ;
73 infile ORD filevar=NextFile dlm=',' end=LastObs;
74 do while (not LastObs);
75 input Customer_ID Order_ID Order_Type
76 Order_Date : date9. Delivery_Date : date9.;
77 output;
78 end;
79 end;
80 stop;
81 run;

NOTE: DATA STEP view saved on file ORION.MOVINGQ.

NOTE: A stored DATA STEP view cannot run under a different operating system.
NOTE: DATA statement used (Total process time):
real time 0.18 seconds
cpu time 0.00 seconds

p310d12
proc print data=orion.movingq;
title 'movingq';
format Order_Date date9.
Delivery_Date date9.;
run;
Partial PROC PRINT Output (created in May)
MovingQ

Customer_ Order_ Order_ Delivery_

Obs ID Order_ID Type Date Date

1 19 1232805509 1 01MAY2007 03MAY2007

2 45 1232857157 2 08MAY2007 12MAY2007
3 908 1232889267 2 13MAY2007 17MAY2007
4 34 1232897220 1 14MAY2007 14MAY2007
5 544 1232936635 2 20MAY2007 21MAY2007
6 111 1232946301 2 22MAY2007 24MAY2007
7 52 1232956741 3 23MAY2007 24MAY2007
8 171 1232972274 1 26MAY2007 26MAY2007
9 183 1232985693 1 28MAY2007 28MAY2007
10 4 1232998740 1 29MAY2007 29MAY2007

(Continued on the next page.)

10-58 Chapter 10 Programmer Efficiency

SAS Log
192 proc print data=orion.movingq;
193 title 'movingq';
194 format Order_Date date9.
195 Delivery_Date date9.;
196 run;

NOTE: The infile ORD is:

Filename=S:\workshop\mon5.dat,
RECFM=V,LRECL=256,File Size (bytes)=411,
Last Modified=27Jan2008:20:32:30,
Create Time=27Jan2008:20:32:30

NOTE: The infile ORD is:

Filename= S:\workshop\\mon4.dat,
RECFM=V,LRECL=256,File Size (bytes)=522,
Last Modified=27Jan2008:20:32:18,
Create Time=27Jan2008:20:32:18

NOTE: The infile ORD is:

Filename= S:\workshop\mon3.dat,
RECFM=V,LRECL=256,File Size (bytes)=334,
Last Modified=27Jan2008:20:32:04,
Create Time=27Jan2008:20:32:04

NOTE: 11 records were read from the infile ORD.

The minimum record length was 34.
The maximum record length was 36.
NOTE: 14 records were read from the infile ORD.
The minimum record length was 34.
The maximum record length was 38.
NOTE: 9 records were read from the infile ORD.
The minimum record length was 34.
The maximum record length was 38.
NOTE: View ORION.MOVINGQ.VIEW used (Total process time):
real time 0.06 seconds
cpu time 0.03 seconds

NOTE: There were 34 observations read from the data set ORION.MOVINGQ.
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.06 seconds
cpu time 0.03 seconds
 10.3 Creating Views 10-59

DATA Statement with VIEW= Option Syntax

General form of the DATA statement with VIEW= option:
DATA data-set-name(s) / VIEW=view-name;
<INFILE fileref;>
<INPUT variable(s);>
<SET/MERGE>
RUN;

view-name specifies a name that the DATA step uses

to store the partially compiled DATA step.

The view-name must match one of the

data set names.

95

You can also create SAS data files in the DATA step that creates the view, but you can only create one
view per DATA step.

 The SAS data file is not created until the view is accessed.

10.09 Quiz
Open and submit the program p310a02.
What does the log report?
data view=orion.movingq;
describe;
run;

97
10-60 Chapter 10 Programmer Efficiency

The DESCRIBE Statement

You can use the DESCRIBE statement to retrieve
program source code from a DATA step view.
SAS writes the source statements to the SAS log.
General form of the DESCRIBE statement:
DATA VIEW =view-name;
DESCRIBE;
RUN;

99
 10.3 Creating Views 10-61

Creating Views with the SQL Procedure

You can create views with the SQL procedure by using
the CREATE VIEW statement.
proc sql;
create view orion.names_view as
select e.Employee_ID, e.Employee_Name,
Manager_ID,
m.Employee_Name as Manager_Name
from orion.staff,
orion.employee_addresses as e,
orion.employee_addresses as m
where e.Employee_ID=staff.Employee_ID
and m.Employee_ID=staff.Manager_ID;
quit;

p310d13
100

The SQL procedure DESCRIBE statement retrieves the SQL view code and reports it in the log.

PROC SQL;
DESCRIBE VIEW view-name;
QUIT;

SAS Log
1213 proc sql;
1214 describe view orion.names_view;
NOTE: SQL view ORION.NAMES_VIEW is defined as:

select e.Employee_ID, e.Employee_Name, Manager_ID, m.Employee_Name as Manager_Name

from ORION.STAFF, ORION.EMPLOYEE_ADDRESSES e, ORION.EMPLOYEE_ADDRESSES m
where (e.Employee_ID=staff.Employee_ID) and (m.Employee_ID=staff.Manager_ID);

1215 quit;
NOTE: PROCEDURE SQL used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds
10-62 Chapter 10 Programmer Efficiency

Creating Views with the SQL Procedure

You can also embed a SAS LIBNAME statement in a
PROC SQL view with the USING clause. This enables
you to store SAS libref information in the view. The scope
of the libref is local to the view, and it will not conflict with
an identically named libref in the SAS session.
proc sql;
create view orion.names_view as
select e.Employee_ID, e.Employee_Name,
Manager_ID,
m.Employee_Name as Manager_Name
from orion.staff,
orion.employee_addresses as e,
orion.employee_addresses as m
where e.Employee_ID=staff.Employee_ID
and m.Employee_ID=staff.Manager_ID
using libname orion 's:\workshop';
quit;
p310d14
101

For more information about the USING clause in PROC SQL, consult SAS OnlineDoc:
SAS OnlineDoc Ö Base SAS Ö SAS 9.2 SQL Procedure User’s Guide Ö
Creating and Updating Tables and Views

Advantages of Views
Advantages of Using Views
Data from multiple sources can be combined.
Complex code can be stored for reuse.
Errors and programming time can be reduced.
You can access the most current data in changing files
A SAS copy of a large data file does not have to be
stored.
You can avoid creating intermediate copies of data.

102
 10.3 Creating Views 10-63

10.10 Quiz
What is the advantage of the following program?
data bonus_view(keep=Manager_ID YrEndBonus)
/ view=bonus_view;
set orion.staff;
YrEndBonus=Salary * 0.05;
where Job_Title contains 'Manager';
run;

proc means data=bonus_view mean sum;

class Manager_ID;
var YrEndBonus;
run;

104 p310d15

Using Views to Avoid Intermediate Data

Views can be used to avoid the I/O associated when you
create a data set to be used in a subsequent step.
Using views to avoid intermediate data has the following
characteristics:
„ does not reduce the CPU resources required to
complete the task. A view might increase the total
CPU time.
„ eliminates writing and reading temporary data sets.

„ reduces the real time required to complete a job

by eliminating one or more I/O bound segments.

106
10-64 Chapter 10 Programmer Efficiency

Using Views to Avoid Intermediate Data

Example of avoiding the I/O needed to create
intermediate data:

data four;
set one two three;
run;
proc sort data=four;
by X;
run;
DATA Step SORT Step

Intermediate Data File

107 continued...

Using Views to Avoid Intermediate Data

Example of avoiding the I/O needed to create
intermediate data:

data four / view=four;

set one two three;
run;
proc sort data=four
out=five;
by X;
run;
DATA Step SORT Step

No Intermediate Data File

108
 10.3 Creating Views 10-65

Using the PRESORTED Option with Views

The following example creates the data view
profit07_view. The SORT step then reads the data from
profit07.dat, validates the sort flag, and creates the data
file profit07.
data profit07_view /
view=profit07_view;
infile 'profit07.dat' dlm=',';
input Company : $30. Sales Cost
Salaries Profit;
run;

proc sort data=profit07_view

out=profit07 presorted;
by Company;
run;

109 p310d16

 PRESORTED is a SAS 9.2 option.

Disadvantages of Views
Disadvantages of Using Views
The code executes each time that you use a view.

System resources are increased.

You run the risk of having your data change between

consecutive executions of the same program.
Depending on how many reads or passes of the data
are required, processing overhead increases.

110
10-66 Chapter 10 Programmer Efficiency

Guidelines for Creating and Using Views

If data is used many times in one program, it is more
efficient to create and reference a SAS data file than
to create and reference a view.

proc print data=orion.sview;

run;

proc freq data=orion.sview;

tables Order_Type;
run;

proc means data=orion.sview;

run;

111

Setup for the Poll

The following program is executed:
data orion.sview / view=orion.sview;
infile 'rawdata.dat';
input variable-list;
<additional statements>
run;
proc print data=orion.sview;
run;
proc freq data=orion.sview;
tables Order_Type;
run;
proc means data=orion.sview;
run;

113
 10.3 Creating Views 10-67

10.11 Multiple Choice Poll

How many times is the rawdata file read?
a. 4
b. 3
c. 2
d. 1
e. 0

114

Guidelines for Creating and Using Views

If data is used many times in one program, it is more
efficient to create and reference a SAS data file than
to create and reference a view.
data order;
set orion.sview;
run;

proc print data=order;

run;

proc freq data=order;

tables Order_Type;
run;

proc means data=order;

run;
116
10-68 Chapter 10 Programmer Efficiency

Guidelines for Creating and Using Views

You might experience a degradation in performance when
you use a SAS data view with a procedure that requires
multiple passes through the data.

proc print data=orion.sview uniform;

run;

117

The PRINT procedure with the UNIFORM option, the CLASS statement in the MEANS/SUMMARY,
TABULATE, and UNIVARIATE procedures, and many SAS/STAT procedures require multiple passes
through the data.

 In the case of multiple passes in a step, the view creates a temporary spill file so that SAS does
not have to read the data from disk multiple times.
 10.3 Creating Views 10-69

Guidelines for Creating and Using Views

In the PRINT procedure, the WIDTH= option along with a
FORMAT statement on variables can reduce execution
time.
proc print data=orion.sview
width=full;
<FORMAT statement to format any
variable that does not have a
permanent format assigned to it>
run;

118

WIDTH=column-width determines the column width for each variable.

value of Use Alias

column-width

FULL uses a variable’s formatted width as the column width.

MINIMUM uses, for each variable, the minimum column width that MIN
accommodates all values of the variable.

UNIFORM uses each variable’s formatted width as its column width on all U
pages. If the variable does not have a format that explicitly
specifies a field width, PROC PRINT uses the widest data value as
the column width.

UNIFORMBY formats all columns uniformly within a BY group, using each UBY
variable’s formatted width as its column width. If the variable does
not have a format that explicitly specifies a field width, PROC
PRINT uses the widest data value as the column width.
10-70 Chapter 10 Programmer Efficiency

Guidelines for Creating and Using Views

Avoid creating views on files whose structures often
change.
filename rawdata 'file1';
file1
1 1 2 proc print data=orion.sview;
1---5----0----5----0 run;
John Smith 21 filename rawdata 'file2'
file2 proc freq data=orion.sview;
1 1 2 tables JobCode;
1---5----0----5----0 run;
John Smith Ted Jones filename rawdata 'file3'
file3 proc means data=orion.sview;
1 1 2 run;
1---5----0----5----0
21 John Smith

119

Which to Use: DATA Step View or SQL View?

DATA Step View SQL View
DATA step views are versatile PROC SQL views do not use DATA
because they use DATA step step programming; it is not as easy
processing, including DO loops and to perform conditional processing.
IF-THEN/ELSE statements.
DATA step views do not have write PROC SQL views can both read
capability; that is, they cannot and update the data that they
directly change the data that they reference. The update can take
access. place only if there is only one table
underlying the view. If there are
joins, you cannot update through an
SQL view.
The DATA step can read data from PROC SQL can only read data that
many different file formats; for is a SAS data set or coming from a
example, text files. relational database.

120 continued...
 10.3 Creating Views 10-71

Which to Use: DATA Step View or SQL View?

DATA Step View SQL View
The DATA step does not support a PROC SQL supports subqueries in
subquery in a WHERE statement. WHERE clauses.
The DATA step does not have an The USING clause in PROC SQL
equivalent to the USING clause in can assign the libref in the view
SQL. itself.
The DATA step can read data from a PROC SQL has a CONNECT TO
DBMS only if a LIBNAME statement component that sends SQL
is assigned using the appropriate statements to a DBMS using the
engine. SQL Pass-Through Facility.

There is no way to subset the The SQL Pass-Through Facility can

underlying data used by a DATA subset your data before processing
step view before using the view. it. This saves memory when you
Even if only part of the data is need to select only a small portion
needed, the entire underlying data of the data referenced in the view.
must be loaded into memory when
the view is used.
121
10-72 Chapter 10 Programmer Efficiency

Reference Information

Creating a View and a File

Only one view can be created in a DATA step.

In addition to the view name, you can specify other data set names in the DATA statement. The data sets
are not created until the view is processed.
p310d17
data orion.movingq errors / view=orion.movingq;
drop MonNum MidMon LastMon i;
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));
do i=MonNum, Midmon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
if _ERROR_=0 then output orion.movingq;
else output errors;
end;
end;
stop;
run;

proc print data=orion.movingq;

title 'movingq';
format Order_Date date9.
Delivery_Date date9.;
run;

proc print data=errors;

title 'Errors';
run;
 10.3 Creating Views 10-73

Using Macro Variables

Because SAS macro variables are resolved during compilation, any macro variables used in a DATA step
view are resolved when the view is created.
You can use the SYMGET function to postpone macro resolution until the view is executed.
p310d18
%let OrderType=2;

data orion.movingq / view= orion.movingq;

drop MonNum MidMon LastMon i Today;
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));
do i=MonNum, Midmon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID
Order_ID
Order_Type
Order_Date : date9.
Delivery_Date : date9.;
if Order_Type=input(symget('OrderType'), 2.) then
output orion.movingq;
end;
end;
stop;
run;

proc print data=orion.movingq;

title 'Using a Macro Variable';
format Order_Date date9.
Delivery_Date date9.;
run;
PROC PRINT Output (from May)
Using a Macro Variable

Customer_ Order_ Order_ Delivery_

Obs ID Order_ID Type Date Date

1 45 1232857157 2 08MAY2007 12MAY2007

2 908 1232889267 2 13MAY2007 17MAY2007
3 544 1232936635 2 20MAY2007 21MAY2007
4 111 1232946301 2 22MAY2007 24MAY2007
5 54655 1232618023 2 03APR2007 06APR2007
6 5 1232728634 2 19APR2007 23APR2007
7 89 1232601472 2 31MAR2007 03APR2007
10-74 Chapter 10 Programmer Efficiency

Exercises

Level 1

4. Creating a DATA Step View

a. Create a view named cc_donations.
1) Read only the observations from the data set orion.employee_donations where the value of
the variable Paid_By is Credit Card.

2) Create a variable named Total_Donations as the total of the variable values for Qtr1, Qtr2,
Qtr3, and Qtr4.
3) Create a new variable Donation_Category with the following values:

Value of Total_Donations Donation_Category

Less than 100 Less than $100

Greater than or equal 100 $100 or more

b. Open and submit the program p310e04 to create a report from the view cc_donations. Use the
variable Donation_Category as a class variable and the variable Total_Donations as an analysis
variable. Verify that the view was created correctly.
p310e04
proc means data=cc_donations sum n nonobs maxdec=2;
class Donation_Category;
var Total_Donations;
run;
Preferred PROC MEANS Output
The MEANS Procedure

Analysis Variable : Total_Donations

Donation_
Category Sum N
ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
$100 or more 200.00 2

Less than $100 1475.00 32

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
 10.3 Creating Views 10-75

Level 2

5. Creating a View and a File in One DATA Step

a. In one DATA step, create a view named younger60 and a file named older60.
b. Read from the data set orion.employee_payroll.
Partial Listing of orion.employee_payroll
Employee_ Birth_ Employee_ Employee_ Marital_
Obs Employee_ID Gender Salary Date Hire_Date Term_Date Status Dependents

1 120101 M 163040 18AUG1976 01JUL2003 . S 0

2 120102 M 108255 11AUG1969 01JUN1989 . O 2
3 120103 M 87975 22JAN1949 01JAN1974 . M 1
4 120104 F 46230 11MAY1954 01JAN1981 . M 1
5 120105 F 27110 21DEC1974 01MAY1999 . S 0

c. Use the variable Birth_Date to calculate each employee's age as of today.

• The view should contain the employees who are younger than 60.
• The file should contain the employees who are 60 or older.
d. Attempt to print the file older60 unsuccessfully.
SAS Log
22 proc print data=older60;
ERROR: File WORK.OLDER60.DATA does not exist.
23 title 'Older60 Data Set';
24 run;

NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.06 seconds
cpu time 0.01 seconds
10-76 Chapter 10 Programmer Efficiency

e. Print the view younger60.

Partial PROC PRINT Output
Younger60 Data Set

Employee_ Birth_ Employee_

Obs Employee_ID Gender Salary Date Hire_Date

1 120101 M 163040 18AUG1976 01JUL2003

2 120102 M 108255 11AUG1969 01JUN1989
3 120104 F 46230 11MAY1954 01JAN1981
4 120105 F 27110 21DEC1974 01MAY1999
5 120108 F 27660 23FEB1984 01AUG2006

Employee_ Marital_
Obs Term_Date Status Dependents Age

1 . S 0 32
2 . O 2 39
3 . M 1 55
4 . S 0 34
5 . S 0 25

 The PROC PRINT output was generated on June 23, 2009. Your results might vary due to
the value of Age.
f. Print the file older60 successfully.
Partial PROC PRINT Output
Older60 Data Set

Employee_ Birth_ Employee_

Obs Employee_ID Gender Salary Date Hire_Date

1 120103 M 87975 22JAN1949 01JAN1974

2 120106 M 26960 23DEC1944 01JAN1974
3 120107 F 30475 21JAN1949 01FEB1974
4 120113 F 26870 10MAY1944 01JAN1974
5 120114 F 31285 08FEB1944 01JAN1974

Employee_ Marital_
Obs Term_Date Status Dependents Age

1 . M 1 60
2 . M 2 64
3 . M 2 60
4 . S 0 65
5 . M 3 65

 The PROC PRINT output was generated on June 23, 2009. Your results might vary due to
the value of Age.
g. Why could you not print older60 in step d?
 10.3 Creating Views 10-77

Level 3

6. Creating a View with the SQL Procedure and the USING Clause
You can embed a SAS LIBNAME statement in a view with the USING clause. This enables you to
store SAS libref information in the view. The scope of the libref is local to the view, and it will not
conflict with an identically named libref in the SAS session.
The starter program p310e06 contains a PROC SQL step that creates a view.
p310e06
proc sql;
create view orion.payroll_donations as
select Employee_ID, Qtr1, Qtr2, Qtr3, Qtr4,
sum(Qtr1, Qtr2, Qtr3, Qtr4) as Total_Donations
from orion.employee_donations
where Paid_By='Payroll Deduction';
quit;

proc print data=orion.payroll_donations;

run;
a. Open the program p310e06, and edit it to assign the libref with the USING clause. The
appropriate library definition is listed below.

Windows s:\workshop

UNIX .

z/OS .prg3.sasdata
b. Submit a LIBNAME statement to assign a libref of sasdata to the library specified in the table
above.
c. Submit a PROC PRINT step to print the view sasdata.payroll_donations.
Partial PROC PRINT Output
orion.payroll_donations View

Total_
Obs Employee_ID Qtr1 Qtr2 Qtr3 Qtr4 Donations

1 120267 15 15 15 15 60
2 120269 20 20 20 20 80
3 120271 20 20 20 20 80
4 120272 10 10 10 10 40
5 120669 15 15 15 15 60

d. Submit a LIBNAME statement to clear the sasdata libref.

10-78 Chapter 10 Programmer Efficiency

10.4 Using FILE and PUT Statements to Create a SAS

Program File

Objectives
„ Use a DATA step to write SAS program code.
„ Include the code and submit it.

125

Using the DATA Step to Create a Program

The DATA step provides the complete control necessary
to write flexible programs.
Selected features when you use the DATA step for writing
SAS code are as follows:
„ FILE statement options

„ line and column pointer controls in a PUT statement

„ IF/THEN logic

„ DATA step functions

„ DO loop processing

126
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-79

Example of Using the DATA Step

proc sql; jobs
create table jobs as Job_Title
select distinct Job_Title Sales Rep. I
from orion.salesstaff; Sales Rep. II
quit; Sales Rep. III
Sales Rep. IV
data _null_;
set jobs;
file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
put 'title "Listing for Job Title '
Job_Title '";';
put 'where Job_Title="' Job_Title
'";' / 'run;' /;
run;

p310d19
128

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV
Job_Title D _N_
Sales Rep. I 1

129 ...
10-80 Chapter 10 Programmer Efficiency

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV
Job_Title D _N_
Sales Rep. I 1

jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5

130 ...

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV
Job_Title D _N_
Sales Rep. I 1

jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5
proc print data=orion.salesstaff;

131 ...
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-81

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV
Job_Title D _N_
Sales Rep. I 1

jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5
proc print data=orion.salesstaff;
title "Listing for Job Title Sales Rep. I";

132 ...

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV
Job_Title D _N_
Sales Rep. I 1

jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5
proc print data=orion.salesstaff;
title "Listing for Job Title Sales Rep. I";
where Job_Title="Sales Rep. I";

133 ...
10-82 Chapter 10 Programmer Efficiency

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV
Job_Title D _N_
Sales Rep. I 1

jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5
proc print data=orion.salesstaff;
title "Listing for Job Title Sales Rep. I";
where Job_Title="Sales Rep. I";
run;

134 ...

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV Implicit RETURN;
Job_Title D _N_
Sales Rep. I 1

jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5
proc print data=orion.salesstaff;
title "Listing for Job Title Sales Rep. I";
where Job_Title="Sales Rep. I";
run;

135 ...
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-83

jobs
data _null_;
Job_Title set jobs;
Sales Rep. I file 'jobs.sas';
put 'proc print data=orion.salesstaff;';
Sales Rep. II put 'title "Listing for Job Title '
Sales Rep. III Job_Title '";';
put 'where Job_Title="' Job_Title
Sales Rep. IV '";' / 'run;' /;
run;
PDV Processing
Job_Title D _N_ continues until
Sales Rep. IV 4 the end of file
in work.jobs.
jobs.sas
1 1 2 2 3 3 4 4 5 5 6 6 7 7
1---5----0----5----0----5----0----5----0----5----0----5----0----5----0----5
proc print data=orion.salesstaff;
title "Listing for Job Title Sales Rep. IV";
where Job_Title="Sales Rep. IV";
run;

136

Using the DATA Step

General form of the DATA step to create a report or raw
data file:

DATA _NULL_;
…DATA step statements…
FILE file-specification;
PUT @n variable1 format … @n variable-n format;
…DATA step statements…
RUN;

Using the _NULL_ option in the DATA statement

does not create a data set when the DATA step
executes.

137
10-84 Chapter 10 Programmer Efficiency

10.12 Multiple Answer Poll

Which resources can you conserve by using the _NULL_
keyword in the DATA statement?
a. I/O
b. CPU
c. Memory
d. Programmer time

139

Using the FILE Statement

Use the FILE statement to specify the external file in
which to write the SAS code generated by the PUT
statements in the current DATA step.
General form of the FILE statement:
FILE file-specification <options>;

The file specification in the FILE statement can be any

one of the following:
„ PRINT to write the PUT statements to the current
output destination
„ the name of an external file

„ a file reference to an external file

Omitting the FILE statement directs the results

141 of the PUT statement to the Log window.
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-85

The %INCLUDE Statement

The %INCLUDE statement retrieves SAS source code
from an external file and compiles and executes the code.
%include 'jobs.sas' / source2;

General form of the %INCLUDE statement:

%INCLUDE file-specification < / SOURCE2 >;

file-specification provides the physical name or fileref

of the file to be retrieved and placed
on the input stack.
SOURCE2 requests that inserted SAS
statements appear in the SAS log.
143

Using the PUT Statement

The PUT statement writes values to an external location
that is specified in the most recently executed FILE
statement.

PUT pointer-controls variables formats 'literals';

Some PUT statement specifications are listed below:

„ constant text

put 'title "Listing for Job Title ' Job_Title '";';

„ variable values
put 'title "Listing for Job Title ' Job_Title '";';

142
10-86 Chapter 10 Programmer Efficiency

Viewing the Source in the Log

77 %include 'jobs.sas' / source2;
NOTE: %INCLUDE (level 1) file jobs.sas is file jobs.sas.
78 +proc print data=orion.salesstaff;
79 +title "Listing for Job Title Sales Rep. I ";
80 +where Job_Title="Sales Rep. I ";
81 +run;

NOTE: There were 63 observations read from the data set ORION.SALESSTAFF.
WHERE Job_Title='Sales Rep. I ';
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

82 +
83 +proc print data=orion.salesstaff;
84 +title "Listing for Job Title Sales Rep. II ";
85 +where Job_Title="Sales Rep. II ";
86 +run;

NOTE: There were 50 observations read from the data set ORION.SALESSTAFF.
WHERE Job_Title='Sales Rep. II ';
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

144 continued...

Viewing the Source in the Log

88 +proc print data=orion.salesstaff;
89 +title "Listing for Job Title Sales Rep. III ";
90 +where Job_Title="Sales Rep. III ";
91 +run;

NOTE: There were 34 observations read from the data set ORION.SALESSTAFF.
WHERE Job_Title='Sales Rep. III ';
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

92 +
93 +proc print data=orion.salesstaff;
94 +title "Listing for Job Title Sales Rep. IV ";
95 +where Job_Title="Sales Rep. IV ";
96 +run;

NOTE: There were 16 observations read from the data set ORION.SALESSTAFF.
WHERE Job_Title='Sales Rep. IV ';
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds

97 +
NOTE: %INCLUDE (level 1) ending.
145
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-87

Using the DATA Step to Send E-Mail

p310d20
You need to send an e-mail that provides the delivery date of a customer’s order. The customer names are
in the data set orion.customer_dim. The delivery date is in the data set orion.order_fact.
In the DATA _NULL_ step, perform the following steps:
1. Merge orion.customer_dim and the sorted copy of orion.order_fact.
2. Use the FILE statement to route the text file to a SAS program file.
3. Create a variable named Address that contains the customer’s e-mail address.
4. Use a PUT statement to create a FILENAME statement with the EMAIL option and the value of
Address, along with a subject for the e-mail.
5. Use PUT statements to write a DATA step that writes the message.
6. Use the %INCLUDE statement to execute the program.

 Do not submit the %INCLUDE statement that is commented out below. There are no mail servers
attached to the classroom machines and the generated e-mail addresses are not valid e-mail
addresses.
p310d20
proc sort data=orion.order_fact(keep=Customer_ID Order_ID
Delivery_Date
obs=50)
out=order_fact;
by Customer_ID;
run;

data _null_;
merge orion.customer_dim(keep=Customer_FirstName
Customer_LastName Customer_ID)
order_fact;
by Customer_ID;
file 'email.sas';
if first.Customer_ID then do;
Address=catt(Customer_FirstName,'.',
Customer_LastName,'@something.com');
FullName=catx(' ', Customer_FirstName, Customer_LastName);
put "filename mail email '" Address "' subject='Purchases';";
put 'data _null_;';
put 'file mail;';
put "put '" FullName +(-1) ",';";
put "put 'Thank you for your orders.';";
put "put 'They will be delivered as follows:'//;";
put "put @10 'Your order number'
@30 'Expected Delivery Date'/;";
(Continued on the next page.)
10-88 Chapter 10 Programmer Efficiency

end;
DT=put(Delivery_Date,mmddyy10.);
put "put @15 '" Order_ID"' @35 '" DT "';";
if last.Customer_ID then do;
put "put /'Your friends at Orion Star';";
put "run;";
end;
run;

/*
%include 'email.sas';
*/
Partial Listing of email.sas
filename mail email '[email protected] ' subject='Purchases';
data _null_;
file mail;
put 'James Kvarniq,';
put 'Thank you for your orders.';
put 'They will be delivered as follows:'//;
put @10 'Your order number' @30 'Expected Delivery Date'/;
put @15 '1232410925 ' @35 '03/03/2004 ';
put @15 '1232455720 ' @35 '03/09/2004 ';
put @15 '1232530384 ' @35 '03/21/2004 ';
put @15 '1232654929 ' @35 '04/09/2004 ';
put @15 '1232654929 ' @35 '04/09/2004 ';
put @15 '1232709099 ' @35 '04/16/2004 ';
put @15 '1232998740 ' @35 '05/29/2004 ';
put @15 '1233543560 ' @35 '08/20/2004 ';
put @15 '1234348668 ' @35 '12/18/2004 ';
put /'Your friends at Orion Star';
run;
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-89

Advantages of Using FILE/PUT to Write Code

The advantanges of using FILE/PUT statements to write
code are as follows:
„ There is more flexibility for generating code than when
using the macro facility.
„ The source code can be kept for future use.

„ You can look at the code before executing it.

„ You can write to several files at the same time. These

files will make up the complete program.
„ The code can be used to write macro code (dynamic
macro names).

147
10-90 Chapter 10 Programmer Efficiency

Exercises

Level 1

7. Creating Two SAS Programs with One DATA Step

The data set orion.customer_type contains the variables Customer_Type_ID, Customer_Type,
Customer_Group_ID, and Customer_Group.
Partial Listing of orion.Customer_Type
Customer_ Customer_
Obs Type_ID Customer_Type Group_ID Customer_Group

1 1010 Orion Club members inactive 10 Orion Club members

2 1020 Orion Club members low activity 10 Orion Club members
3 1030 Orion Club members medium activity 10 Orion Club members
4 1040 Orion Club members high activity 10 Orion Club members
5 2010 Orion Club Gold members low activity 20 Orion Club Gold members
6 2020 Orion Club Gold members medium activity 20 Orion Club Gold members
7 2030 Orion Club Gold members high activity 20 Orion Club Gold members
8 3010 Internet/Catalog Customers 30 Internet/Catalog Customers

a. Write a DATA step to build the following two PROC FORMAT steps. Under Windows and UNIX,
name the files customer_type.sas and customer_group.sas. Under z/OS, name the files
.workshop.sascode(customer_type) and .workshop.sascode(customer_group).
Preferred Output from the DATA Step for customer_group
proc format fmtlib;
value GrpLevl
10="Orion Club members"
20="Orion Club Gold members"
30="Internet/Catalog Customers";
run;
Preferred Output from the DATA Step for customer_type
proc format fmtlib;
value TypeLevl
1010="Orion Club members inactive"
1020="Orion Club members low activity"
1030="Orion Club members medium activity"
1040="Orion Club members high activity"
2010="Orion Club Gold members low activity"
2020="Orion Club Gold members medium activity"
2030="Orion Club Gold members high activity"
3010="Internet/Catalog Customers";
run;
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-91

Hint: To create the value for the label, use the $QUOTE format that writes data values that are
enclosed in double quotation marks. Investigate the descriptor portion of the data set to
determine the appropriate width.
b. Use the %INCLUDE statement to execute the code.
Partial SAS Log
95 %include 'Customer_Type.sas'/source2;
NOTE: %INCLUDE (level 1) file Customer_Type.sas is file S:\workshop\Customer_Type.sas.
96 +proc format fmtlib;
97 +value TypLevl
98 +1010="Orion Club members inactive"
99 +1020="Orion Club members low activity"
100 +1030="Orion Club members medium activity"
101 +1040="Orion Club members high activity"
102 +2010="Orion Club Gold members low activity"
103 +2020="Orion Club Gold members medium activity"
104 +2030="Orion Club Gold members high activity"
105 +3010=Internet/Catalog Customers"
106 +;
NOTE: Format TYPLEVL has been output.
106!+ run;

NOTE: PROCEDURE FORMAT used (Total process time):

real time 0.23 seconds
cpu time 0.04 seconds

Level 2

8. Sending E-Mail Using the DATA Step

The data set orion.employee_donations contains the variables Employee_ID, Qtr1, Qtr2, Qtr3,
Qtr4, and Recipients.
Partial Listing of orion.Employee_Donations
Employee
_ID Qtr1 Qtr2 Qtr3 Qtr4 Recipients

120265 . . . 25 Mitleid International 90%,

Save the Baby Animals 10%
120267 15 15 15 15 Disaster Assist, Inc. 80%,
Cancer Cures, Inc. 20%
120269 20 20 20 20 Cancer Cures, Inc. 10%,
Cuidadores Ltd. 90%
120270 20 10 5 . AquaMissions International 10%,
Child Survivors 90%
120271 20 20 20 20 Cuidadores Ltd. 80%,
Mitleid International 20%

The data set orion.employee_addresses contains the names of the employees.

10-92 Chapter 10 Programmer Efficiency

Open the program p310e08, which contains a DATA step with a MERGE statement, and edit the
program to generate an e-mail for each employee to inform him that the total contribution was mailed.
Under Windows and UNIX, name the file donations.sas. Under z/OS, name the file
.workshop.sascode(donations). Do not include the program file. However, open the program file to
verify that it is correct.
p310e08
proc sort data=orion.employee_addresses out=employee_addresses;
by Employee_ID;
run;

data _null_;
merge orion.employee_donations(in=d) employee_addresses;
by Employee_ID;
if d;
run;
Partial Contents of donations
filename mail email '[email protected] ' subject='Your Donation';
data _null_;
file mail;
put 'Your donation of $25 has been sent to Mitleid International
90%, Save the Baby Animals 10% ';
run;
filename mail email '[email protected] ' subject='Your Donation';
data _null_;
file mail;
put 'Your donation of $60 has been sent to Disaster Assist, Inc.
80%, Cancer Cures, Inc. 20% ';
run;
 10.4 Using FILE and PUT Statements to Create a SAS Program File 10-93

Level 3

9. Writing PROC PRINT Steps Using SQL DICTIONARY Table Views

You can use the SQL procedure to obtain information about your SAS session. This information is
stored in special tables called DICTIONARY tables that are only available in an SQL step. However,
there are PROC SQL views, stored in the Sashelp library, that reference the DICTIONARY tables and
that can be used in other SAS procedures and in the DATA step.
One of those views, Sashelp.VCOLUMN, contains information such as name, type, length, and
format about all columns in all tables contained in the currently assigned librefs.
a. Use the DATA step to read from Sashelp.VCOLUMN and create a program containing PROC
PRINT steps that create reports for every data set in the orion library that contains the variable
Product_ID. PROC PRINT should only print five observations from each of the data sets.
Hint: The variables in Sashelp.VCOLUMN that you need to use are as follows:

Name of Variable Value of Variable Use in the Program

LIBNAME ORION In a WHERE statement

COLUMN Product_ID In a WHERE statement

MEMNAME The value you need to retrieve In the DATA= option in the PROC PRINT
from Sashelp.VCOLUMN statement and in the TITLE statement

 The variable values in Sashelp.VCOLUMN are case sensitive.

The starter program p310e09 can be used as a starting point for this exercise.
p310e09
data _null_;
set sashelp.vcolumn;
where Libname='ORION' and Name='Product_ID';
file 'print_products.sas';
run;
10-94 Chapter 10 Programmer Efficiency

b. Store the program in a file named Print_Products.sas under Windows and UNIX and
.workshop.sascode(Print_Products) under z/OS.
Partial Contents of Print_Products
proc print data=ORION.CATALOG(obs=5);run;
title 'First Five Observations of ORION.CATALOG ';
run;
proc print data=ORION.DENMARK_CUSTOMERS(obs=5);run;
title 'First Five Observations of ORION.DENMARK_CUSTOMERS ';
run;
proc print data=ORION.FIRST_INTERNET_ORDER(obs=5);run;
title 'First Five Observations of ORION.FIRST_INTERNET_ORDER ';
run;
proc print data=ORION.INTERNET(obs=5);run;
title 'First Five Observations of ORION.INTERNET ';
run;
proc print data=ORION.MULTIPLE_ORDERS(obs=5);run;
title 'First Five Observations of ORION.MULTIPLE_ORDERS ';
run;
proc print data=ORION.NEW_PRODUCTS(obs=5);run;
title 'First Five Observations of ORION.NEW_PRODUCTS ';
run;
proc print data=ORION.ORDER_FACT(obs=5);run;
title 'First Five Observations of ORION.ORDER_FACT ';
run;
proc print data=ORION.PRICE_LIST(obs=5);run;
title 'First Five Observations of ORION.PRICE_LIST ';
run;
proc print data=ORION.PRODUCT_DIM(obs=5);run;
title 'First Five Observations of ORION.PRODUCT_DIM ';
run;
proc print data=ORION.PRODUCT_LIST(obs=5);run;
title 'First Five Observations of ORION.PRODUCT_LIST ';
run;
c. Use the %INCLUDE statement to execute the code or open the program to verify that it is correct.
(You can use the SAS Editor window, PROC FSLIST, or Notepad to verify the contents of the
program file.)
 10.5 Using the FCMP Procedure (Self-Study) 10-95

10.5 Using the FCMP Procedure (Self-Study)

Objectives
„ List reasons to use the FCMP procedure.
„ Examine the syntax for the FCMP procedure.
„ Create functions using the FCMP procedure.
„ Use the user-written functions.
„ Create subroutines using the FCMP procedure.

151

The FCMP procedure is new for use in the DATA step in SAS 9.2.

Using PROC FCMP

PROC FCMP is a method of simplifying the repetitive use
of business rules. PROC FCMP provides the following
functionality:
„ ability to write functions and CALL routines using
DATA step syntax
„ storage of these functions in a package within
a SAS data set
„ use of functions and subroutines previously defined in
the current FCMP procedure step, as well as most
DATA step functions, within the routines

 In SAS 9.2, these functions can be called

from a DATA step.
152
10-96 Chapter 10 Programmer Efficiency

Why Create a Function with PROC FCMP?

PROC FCMP creates programs that have the following
attributes:
„ simpler programs that are easier to read, write,
and modify
„ independent routines because any program that
calls a routine is not affected by the routine's
implementation
„ reusable routines that can be called by any program
that can read the data set where the routine is stored

153

Business Scenario
The Marketing Department at Orion needs to have a
report created daily. The requirements are that the report
must include a column that is the customer ID
concatenated with a comment.

Partial Listing
Using the FCMP Procedure

Delivery_ Order_
Obs Customer_ID Date Type Marketing_Comment

1 63 11JAN2003 1 000000000063 - Mail In-Store Coupon

2 5 19JAN2003 2 000000000005 - Send New Catalog
3 45 22JAN2003 2 000000000045 - Send New Catalog
4 41 28JAN2003 1 000000000041 - Mail In-Store Coupon
5 183 27FEB2003 1 000000000183 - Mail In-Store Coupon

154
 10.5 Using the FCMP Procedure (Self-Study) 10-97

Example of PROC FCMP

proc fcmp outlib=orion.functions.Marketing;
function MKT(ID, Date, Type) $ 40;
if '01Jan2008'd - Date>90 then do;
if Type=1 then return(catx(' - ',
put(ID, z12.), 'Mail In-Store Coupon'));
else if Type=2 then return(catx(' - ',
put(ID, z12.), 'Send New Catalog'));
else return(catx(' - ',
put(ID, z12.),'Send Email'));
end;
else return(catx(' - ', put(ID, z12.),
'Wait to Contact'));
endsub;
run;
quit;

p310d21
155

The PROC FCMP Statement

proc fcmp outlib=orion.functions.Marketing;

General form of the PROC FCMP statement:

PROC FCMP OUTLIB= | OUTCAT=libname.data-set.package;

OUTLIB= | OUTCAT= specifies the three-level name

libname.data- of an output package to which
set.package the compiled subroutines and
functions are written when the
PROC FCMP step ends.

The OUTLIB= | OUTCAT= argument is required.

156
10-98 Chapter 10 Programmer Efficiency

The FUNCTION Statement

proc fcmp outlib=orion.functions.Marketing;
function MKT(ID, Date, Type) $ 40;

„ This code creates a function named MKT in a package

named Marketing.
„ The package is stored in the data set orion.functions.

 A package is a collection of routines that have

unique names.

157

The Function Definition

Within one PROC FCMP step, you can define multiple
functions.
Each function definition consists of this block of code:

FUNCTION name (parameter-1, …,

parameter-N);
program-statements;
RETURN (expression);
ENDSUB;

158
 10.5 Using the FCMP Procedure (Self-Study) 10-99

The FUNCTION Statement

function MKT(ID, Date, Type) $ 40;

General form of the FUNCTION statement:

FUNCTION function-name(argument-1 <$>,
argument-2 <$>, ...,
argument-n <$>) <$> <length>;

function-name specifies the name of the function.

argument specifies one or more arguments in the
function.
$ specifies a character value.
length specifies the length of a character value.
159

The RETURN Statement

if '01Jan2008'd – Date > 90 then do;
if Type=1 then return(catx(' - ',
put(ID, z12.), 'Mail In-Store Coupon'));
else if Type=2 then return(catx(' - ',
put(ID, z12.), 'Send New Catalog'));
else return(catx(' - ',
put(ID, z12.), 'Send Email'));
end;
else return(catx(' - ', put(ID, z12.),
'Wait to Contact'));

General form of the RETURN statement:

RETURN (expression);

expression specifies the value that is returned from

the function.

160
All functions must return a value.
10-100 Chapter 10 Programmer Efficiency

The ENDSUB Statement

The ENDSUB statement ends the function definition.
endsub;

General form of the ENDSUB statement:

ENDSUB;

161

Using the Function

The functions created by PROC FCMP can be called
within the following:
„ the DATA step

„ the COMPUTE block within a REPORT procedure

step
„ the SQL procedure

„ selected SAS/STAT procedures

„ selected SAS/ETS procedures

„ SAS/OR procedures, such as the NLIN, MODEL,

and NLP procedures

162

 Support for PROC FCMP functions used in WHERE statements and PROC COMPUTAB was
added in the platform for SAS Business Analytics 9.2 release.
 10.5 Using the FCMP Procedure (Self-Study) 10-101

Using a Function
options cmplib=orion.functions;

data temp;
set orion.order_fact;
Marketing_Comment=
MKT(Customer_ID,Delivery_Date,Order_Type);
run;

proc print data=temp(obs=5);

title 'Using the FCMP Procedure';
var Customer_ID Delivery_Date
Order_Type Marketing_Comment;
run;

p310d22
163

Using the FCMP Function

Partial Listing
Using the FCMP Procedure

Delivery_ Order_
Obs Customer_ID Date Type Marketing_Comment

1 63 11JAN2003 1 000000000063 - Mail In-Store Coupon

2 5 19JAN2003 2 000000000005 - Send New Catalog
3 45 22JAN2003 2 000000000045 - Send New Catalog
4 41 28JAN2003 1 000000000041 - Mail In-Store Coupon
5 183 27FEB2003 1 000000000183 - Mail In-Store Coupon

164
10-102 Chapter 10 Programmer Efficiency

Using the Function

In order to use the function in a DATA step or supported
PROC step, use the CMPLIB= SAS system option.
options cmplib=orion.functions;

General form of the CMPLIB= SAS system option:

CMPLIB=libref.data-set | (libref. data-set-1 ... libref. data-set-n);

libref. data-set specifies the libref and data set name

of the compiled subroutines that are to
be included during the program
compilation.

165

 The order of the libref.data-set names in the list (libref. data-set-1 ... libref. data-set-n)
determines the order in which the data sets are searched.

Setup for the Quiz

proc fcmp outlib=orion.functions.Marketing;
function MKT(ID, Date, Type) $ 40;
if '01Jan2008'd – Date > 90 then do;
if Type=1 then return(catx(' - ',
put(ID, z12.), 'Mail In-Store Coupon'));
else if Type=2 then return(catx(' - ',
put(ID, z12.), 'Send New Catalog'));
else return(catx(' - ',
put(ID, z12.), 'Send Email'));
end;
else return(catx(' - ', put(ID, z12.),
'Wait to Contact'));
endsub;
run;
quit;

167
 10.5 Using the FCMP Procedure (Self-Study) 10-103

10.13 Quiz
Specify the argument of the MKT function that
corresponds to the each of following variables:
Variable Argument
Customer_ID
Delivery_Date
Order_Type
options cmplib=orion.functions;
data temp;
set orion.order_fact;
Marketing_Comment =
MKT(Customer_ID,Delivery_Date,Order_Type);
run;
168

Business Scenario
You need to create two functions.
Function Name Use of Function

MONDAY takes a SAS date as the argument and creates

a SAS date for the Monday of that week.

FRIDAY takes a SAS date as the argument and creates

a SAS date for the Friday of that week.

proc fcmp outlib=orion.functions.DateType;

function MONDAY(Date);
return(intnx('week.2', Date, 0));
endsub;
function FRIDAY(Date);
return(intnx('week.7', Date, 1)- 1);
endsub;
run;
quit;
170 p310d23
10-104 Chapter 10 Programmer Efficiency

Using the INTNX Function

You can create multiples of intervals and shift their
starting points to construct more complex interval
specifications by using multipliers and shift indexes in the
INTNX function.
General form of the INTNX function:

INTNX(interval<multiple><.shift-index>, start-from, increment<,alignment>)

interval<multiple><.shift-index>

171

Using the INTNX Function

date='01jan2008'd; Tuesday, January 1, 2008

Function Result
INTNX('week',Date,0) Sunday, December 30, 2007

INTNX('week.2',Date,0) Monday, December 31, 2007

INTNX('week2',Date,0) Sunday, December 23, 2007

INTNX('week2.2',Date,0) Monday, December 24, 2007

INTNX('week.6',Date,1) Friday, January 4, 2008

INTNX('week.7',Date,1) Saturday, January 5, 2008

INTNX('week.7',Date,1)-1 Friday, January 4, 2008

172
 10.5 Using the FCMP Procedure (Self-Study) 10-105

Business Scenario
The data set orion.Order_Fact contains the delivery date
and order number for customer orders. If the delivery date
is on Saturday or Sunday, the order will be delivered on
Saturday. However, if the delivery date is a weekday, then
the order will be delivered on some day between Monday
and Friday in that week.
Partial Listing
Delivery Information

Customer Order Number When you can expect your delivery

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ
4 1232410925 Between Monday, March 1, 2004 and Friday, March 5, 2004
1232455720 Between Monday, March 8, 2004 and Friday, March 12, 2004
1232530384 Saturday, March 20, 2004
1232654929 Between Monday, April 5, 2004 and Friday, April 9, 2004
1232709099 Between Monday, April 12, 2004 and Friday, April 16, 2004
1232998740 Saturday, May 29, 2004
1233543560 Between Monday, August 16, 2004 and Friday, August 20, 2004
1234348668 Saturday, December 18, 2004

5 1230080101 Saturday, January 18, 2003

173
10-106 Chapter 10 Programmer Efficiency

Creating and Using Functions

p310d23
proc fcmp outlib=orion.functions.DateType;
function MONDAY(Date);
return(intnx('week.2', Date, 0));
endsub;
function FRIDAY(Date);
return(intnx('week.7', Date, 1)-1);
endsub;
run;
quit;

option cmplib=orion.functions;

proc report data=orion.Order_Fact ls=120

headline headskip nowd;
title 'Delivery Information';
column Customer_ID Order_ID Delivery_Date
Expect_Delivery;
define Customer_ID / group width =10 'Customer';
define Delivery_Date / group noprint;
define Order_ID / group 'Order Number' width=12;
define Expect_Delivery / computed
'When you can expect your delivery'
width=70 center;
compute Expect_Delivery / char length=70;
First_Of_Week=MONDAY(Delivery_Date);
End_Of_Week=FRIDAY(Delivery_Date);
if weekday(Delivery_Date) not in (1, 7) then
Expect_Delivery=catx(' ', 'Between',
put(First_Of_Week, weekdate.),
'and', put(End_Of_Week, weekdate.));
else if weekday(Delivery_Date)=1 then
Expect_Delivery=put(Delivery_Date - 1, weekdate.);
else Expect_Delivery=put(Delivery_Date, weekdate.);
endcomp;
break after Customer_ID / skip;
run;
 10.5 Using the FCMP Procedure (Self-Study) 10-107

Partial Listing
Delivery Information

Customer Order Number When you can expect your delivery

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ

4 1232410925 Between Monday, March 1, 2004 and Friday, March 5, 2004

1232455720 Between Monday, March 8, 2004 and Friday, March 12, 2004
1232530384 Saturday, March 20, 2004
1232654929 Between Monday, April 5, 2004 and Friday, April 9, 2004
1232709099 Between Monday, April 12, 2004 and Friday, April 16, 2004
1232998740 Saturday, May 29, 2004
1233543560 Between Monday, August 16, 2004 and Friday, August 20, 2004
1234348668 Saturday, December 18, 2004

5 1230080101 Saturday, January 18, 2003

1231663230 Between Monday, October 27, 2003 and Friday, October 31, 2003
1231842118 Between Monday, December 1, 2003 and Friday, December 5, 2003
1231950921 Between Monday, December 22, 2003 and Friday, December 26, 2003
1231956902 Between Monday, December 22, 2003 and Friday, December 26, 2003
1232007693 Saturday, January 3, 2004
1232728634 Between Monday, April 19, 2004 and Friday, April 23, 2004
1233315988 Between Monday, July 12, 2004 and Friday, July 16, 2004
1233682051 Between Monday, September 13, 2004 and Friday, September 17, 2004
1237890730 Between Monday, December 5, 2005 and Friday, December 9, 2005
1242140006 Between Monday, May 7, 2007 and Friday, May 11, 2007
1242159212 Between Monday, May 7, 2007 and Friday, May 11, 2007
1242493791 Saturday, June 9, 2007
1243315613 Saturday, September 8, 2007
1244296274 Between Monday, December 24, 2007 and Friday, December 28, 2007

9 1232698281 Between Monday, April 19, 2004 and Friday, April 23, 2004
1236028541 Saturday, June 11, 2005
1236673732 Between Monday, August 15, 2005 and Friday, August 19, 2005
1237825036 Between Monday, December 5, 2005 and Friday, December 9, 2005
1238053337 Between Monday, December 26, 2005 and Friday, December 30, 2005
10-108 Chapter 10 Programmer Efficiency

Advantages of PROC FCMP

Advantages of Using the FCMP Procedure
A library of reusable routines can be built using DATA
step syntax.

Complex programs can be simplified by abstracting

common computations into named program units.

FCMP routines are independent from their use and can

be reused in any DATA step program that has access to
their storage locations.
FCMP routines enable programmers to read, write, and
maintain complex code easily.

175

Disadvantages of PROC FCMP

Disadvantages of Using the FCMP Procedure
User-written functions and routines can only be used in
the DATA step, COMPUTE blocks in PROC REPORT,
PROC SQL, and selected procedures in SAS/STAT,
SAS/ETS, and SAS/OR.
Knowledge of the FCMP syntax is needed.

176
 10.5 Using the FCMP Procedure (Self-Study) 10-109

Creating Subroutines Using PROC FCMP

To create a subroutine using the FCMP procedure,
use the following syntax:
SUBROUTINE subroutine-name (argument-1,..., argument-n);
OUTARGS out-argument-1, ..., out-argument-n;
... more-program-statements ...
ENDSUB;

subroutine-name specifies the name of a subroutine.

argument specifies one or more arguments in

the subroutine.
out-argument specifies arguments from the
argument list that you want the
subroutine to update.
177

 Support for PROC FCMP subroutines used in %SYSFUNC and %SYSCALL macro functions,
ODS tagsets, and the Graph Template Language was added in the platform for SAS Business
Analytics 9.2 release.

Difference between Functions

and CALL Routines
Functions Call Routines
Can require arguments and Do not return a value and can
must return a value. modify their parameters.

All parameters are passed by The formal parameters listed in

value. This means that the value the OUTARGS statement are
of the actual parameter, the passed by reference instead of
variable, or the value passed to by value. This means that any
the function from the calling modification of the formal
environment is copied before parameter by the CALL routine
being used by the function. This modifies the original variable
ensures that any modification of that was passed.
the formal parameter by the
function does not change the
original value.
178
10-110 Chapter 10 Programmer Efficiency

Creating Subroutines Using PROC FCMP

p310d24
proc fcmp outlib=orion.functions.Directory;
function DIROPEN(DIR$);
length DIR$ 256 FREF $ 8;
rc=filename(FREF, DIR);
if rc=0 then do;
DID=dopen(FREF);
rc=filename(FREF);
end;
else do;
MSG=sysMSG();
put MSG '(DIROPEN(' DIR= ')';
DID=.;
end;
return(DID);
endsub;

subroutine DIRCLOSE(DID);
outargs DID;
rc=dCLOSE(DID);
DID=.;
endsub;

subroutine DIR_entries(DIR$, FILES[*] $, N, TRUNC);

outargs FILES, N, TRUNC;
length Entry $ 256;
if TRUNC then return;
DID=DIROPEN(DIR);
if DID <= 0 then return;
dnum=dnum(DID);
do i=1 to DNUM;
ENTRY=dread(DID, i);
(Continued on the next page.)
 10.5 Using the FCMP Procedure (Self-Study) 10-111

/* If this entry is a file, then add to array */

/* Else ENTRY is a directory, recurse. */
FID=mopen(DID, ENTRY);
ENTRY=trim(DIR) || '\' || ENTRY;
if FID > 0 then do;
rc=fCLOSE(FID);
if n < dim(FILES) then do;
TRUNC=0;
N=N + 1;
FILES{N}=ENTRY;
end;
else do;
DID=1;
call DIRCLOSE(DID);
return;
end;
end;
else
call DIR_entries(ENTRY, FILES, N, TRUNC);
end;
call DIRCLOSE(DID);
return;
endsub;
run;

options cmplib=orion.functions;

data _null_;
array FILES[1000] $ 256 _temporary_;
DNUM=0;
TRUNC=0;
call DIR_entries(".", FILES, DNUM, TRUNC);
if TRUNC then put 'ERROR: Not enough result array entries.

increase array size.';

do i=1 to DNUM;
put FILES{i};
end;
run;
10-112 Chapter 10 Programmer Efficiency

Partial SAS Log

70 options cmplib=orion.functions;
71 data _null_;
72 array files[1000] $ 256 _temporary_;
73 DNUM=0;
74 TRUNC=0;
75 call dir_entries("S:\Workshop\", FILES, DNUM,
75 ! trunc);
76 if trunc then put 'ERROR: Not enough result array entries.
76 ! Increase array size.';
77 do i=1 to DNUM;
78 put files[i];
79 end;
80 run;

S:\Workshop\city.sas7bdat
S:\Workshop\continent.sas7bdat
S:\Workshop\country.sas7bdat
S:\Workshop\county.sas7bdat
S:\Workshop\customer.sas7bdat
S:\Workshop\customer_dim.sas7bdat
S:\Workshop\customer_type.sas7bdat
S:\Workshop\discount.sas7bdat
S:\Workshop\employee_addresses.sas7bdat
S:\Workshop\employee_organization.sas7bdat
S:\Workshop\employee_payroll.sas7bdat
S:\Workshop\employee_phones.sas7bdat
S:\Workshop\funcs.sas7bdat
S:\Workshop\funcs.sas7bndx
S:\Workshop\functions.sas7bdat
S:\Workshop\functions.sas7bndx
S:\Workshop\geography_dim.sas7bdat
S:\Workshop\geo_type.sas7bdat
S:\Workshop\lookup_agegroup.sas7bdat
S:\Workshop\lookup_country.sas7bdat
S:\Workshop\lookup_custgrp.sas7bdat
S:\Workshop\lookup_euday.sas7bdat
S:\Workshop\lookup_order_type.sas7bdat
S:\Workshop\lookup_product.sas7bdat
S:\Workshop\lookup_usday.sas7bdat
S:\Workshop\orders.sas7bdat
S:\Workshop\order_fact.sas7bdat
S:\Workshop\order_item.sas7bdat
S:\Workshop\organization.sas7bdat
S:\Workshop\organization_dim.sas7bdat
S:\Workshop\org_level.sas7bdat
S:\Workshop\postal_code.sas7bdat
S:\Workshop\price_list.sas7bdat
S:\Workshop\product_dim.sas7bdat
S:\Workshop\product_level.sas7bdat
S:\Workshop\product_list.sas7bdat
S:\Workshop\staff.sas7bdat
S:\Workshop\state.sas7bdat
S:\Workshop\street_code.sas7bdat
S:\Workshop\supplier.sas7bdat
S:\Workshop\time_dim.sas7bdat

 Your results might show a different path.

 10.5 Using the FCMP Procedure (Self-Study) 10-113

DATA Step Functions Used in the

Demonstration
filename assigns or de-assigns a fileref to an external file, directory,
or output device.
dopen opens a directory using a previously assigned fileref and
returns a directory identifier value.
sysmsg returns the text of error messages or warning messages
from the last data set or external file function executed.
dclose closes a directory that was opened by the DOPEN function.

dnum returns the number of members in a directory.

dread returns the name of a directory member.

mopen opens a file by directory ID and member name, and returns

the file identifier or a 0.
fclose closes an external file, directory, or directory member.
180

Using Arrays in PROC FCMP

You can use arrays in PROC FCMP.
„ The ARRAY statement in PROC FCMP is similar to
the ARRAY statement that is used in the DATA step.
„ In FCMP routines, arrays can be resized. This is done
by calling the built-in CALL routine DYNAMIC_ARRAY.

When you reference an array, square braces [ ] or

curly braces { } must be used.

181
10-114 Chapter 10 Programmer Efficiency

The Scope of Argument Variables in

User-Defined Functions
The scope of argument variables in user-defined
functions is as follows:
„ A variable's scope is the section of code where
a variable's value can be used.
„ Variables declared outside of routines created by
PROC FCMP are not accessible inside a routine.
„ Variables declared inside a routine are not accessible
outside of the routine. Variables declared within a
routine are called local variables because their scope
is “local” to the routine.
Local variables store intermediate results of a
computation and cannot be accessed after a routine
returns.
182
 10.5 Using the FCMP Procedure (Self-Study) 10-115

Exercises

Level 1

10. Using the FCMP Procedure to Store a Formula in a Function

a. Open the program p310e10 and submit it.
p310e10
data test;
set orion.order_fact(keep=Employee_ID Quantity
Total_Retail_Price);
if Quantity > 2 then
Kick_Back_Amt=Quantity * Total_Retail_Price / 5;
else Kick_Back_Amt=Quantity * Total_Retail_Price / 10;
run;

proc print data=test(obs=5);

run;
b. Use PROC FCMP to encapsulate the IF/THEN logic into a function named KB. Store the
function in work.functions.Marketing.
c. Write a DATA step to create a data set named kick_backs that uses the KB function to create a
variable named Kick_Back_Amt. Set the appropriate system option so that the new function can
be utilized. The DATA step should not contain any IF/THEN logic.
d. Print the first five observations of the SAS data set kick_backs.
PROC PRINT Output
Total_Retail_ Kick_
Obs Employee_ID Quantity Price Back_Amt

1 121039 1 $16.50 1.65

2 99999999 1 $247.50 24.75
3 99999999 1 $28.30 2.83
4 120174 2 $32.00 6.40
5 120134 3 $63.60 38.16
10-116 Chapter 10 Programmer Efficiency

Level 2

11. Using the FCMP Procedure to Store a Date Calculation in a Function

The INTCK function can be used to calculate the differences between two dates, but it does not take
into consideration whether either of the dates is a leap year date (February 29). If that is important to
your date calculations, the following formula can be used. (An explanation of the formula is at the
end of the exercise description.)
intck('year',BirthDate,ActualDate)
-(put(BirthDate,mmddyy4.) gt put(ActualDate,mmddyy4.))
+(put(BirthDate,mmddyy4.)||put(ActualDate,mmddyy4.)||
put(ActualDate+1,mmddyy4.)='022902280301')
a. Open the program p310e11 that contains the formula and a DATA step that generates test data.
p310e11
data real_ages;
do Birth_Date='28feb1960'd to '01mar1960'd;
do Actual_Date='28feb2004'd to '01mar2004'd,
'28feb2005'd to '01mar2005'd;

/* Calculate Real Age */

Age=intck('year', Birth_Date, Actual_Date);

output;
end;
end;
format Birth_Date Actual_Date worddate.;
run;

proc print data=real_ages;

var Birth_Date Actual_Date Age;
title1 'Age Calculations based using INTCK';
run;
b. Write a PROC FCMP step that creates a function named AGE that contains the formula. Store the
function in work.functions.Marketing.
c. To the DATA step in p310e11, add an assignment statement that creates a variable named
Real_Age using the AGE function and a variable named Age using the INTCK function.
 10.5 Using the FCMP Procedure (Self-Study) 10-117

d. Print the data to ensure that the function is correctly calculating Real_Age.
PROC PRINT Output
Age Calculations based using INTCK
Obs Birth_Date Actual_Date Real_Age Age

1 February 28, 1960 February 28, 2004 44 44

2 February 28, 1960 February 29, 2004 44 44
3 February 28, 1960 March 1, 2004 44 44
4 February 28, 1960 February 28, 2005 45 45
5 February 28, 1960 March 1, 2005 45 45
6 February 29, 1960 February 28, 2004 43 44
7 February 29, 1960 February 29, 2004 44 44
8 February 29, 1960 March 1, 2004 44 44
9 February 29, 1960 February 28, 2005 45 45
10 February 29, 1960 March 1, 2005 45 45
11 March 1, 1960 February 28, 2004 43 44
12 March 1, 1960 February 29, 2004 43 44
13 March 1, 1960 March 1, 2004 44 44
14 March 1, 1960 February 28, 2005 44 45
15 March 1, 1960 March 1, 2005 45 45

e. Create a data set named customer_ages from the orion.customer_dim data set. The new data set
should contain a new variable named Real_Age using the AGE function with
Customer_BirthDate as the Birth_Date variable and 01JAN2008 as the value of Actual_Date.
The new data set should also contain a new variable named Age calculated using the INTCK
function. Print the first five observations of customer_ages.

 There is a variable, Customer_Age, in the data set orion.customer_dim. Do not use this
variable.
PROC PRINT Output
Age Calculations based on Calendar-Based Algorithm

Customer_
Obs Customer_ID Customer_Name BirthDate

1 4 James Kvarniq June 27, 1974

2 5 Sandrina Stephano July 9, 1979
3 9 Cornelia Krahl February 27, 1974
4 10 Karen Ballinger October 18, 1984
5 11 Elke Wallstab August 16, 1974

Obs Customer_Group Real_Age Age

1 Orion Club members 33 34

2 Orion Club Gold members 28 29
3 Orion Club Gold members 33 34
4 Orion Club members 23 24
5 Orion Club members 33 34
10-118 Chapter 10 Programmer Efficiency

Explanation of the Formula Used in This Exercise

The following describes the syntax when BirthDate='29FEB2008'D and
ActualDate='28FEB2009'd:
• intck('year', BirthDate, ActualDate) returns the number of January firsts (1st)
between the two dates. In this example, there is one January 1st between the two dates.
• put(BirthDate, mmddyy4.) gt put(ActualDate, mmddyy4.) determines
whether the birthday is later than the actual day. If the expression is true, it returns a 1; if false,
a 0. Because 0229 > 0203, the expression returns a 1. In this example, 0229 gt 0228 is
true, so 1 is subtracted from the number of years returned from the INTCK formula.
• put(BirthDate,mmddyy4.)||put(ActualDate,mmddyy4.)||
put(ActualDate+1,mmddyy4.)='022902280301' compares the text value
022902280301 to 022902280301. If the expression is true, it returns a 1; if false, a 0. In this
case, the expression is true, so 1 is added to the number of years returned from the INTCK
formula.

Level 3

12. Creating a Function Using DATA SAS File I/O Functions

a. Use PROC FCMP to create a function named NUMS that takes one argument, which is the name
of a SAS data set, and returns the number of logical observations in the data set. Refer to SAS
OnlineDoc or the SAS Help facility to determine which SAS file I/O functions are required.
b. Open the program p310e12 and submit it. Look in the log to ensure that the function is working
correctly.
p310e12
data _null_;
X=NUMS('orion.internet');
put X=;
run;
Partial SAS Log
218 data _null_;
219 X=NUMS('orion.Internet');
220 put X=;
221 run;
X=123
 10.6 Chapter Review 10-119

10.6 Chapter Review

Chapter Review
1. What are two parts of the macro facility?

2. What is the purpose of the FILEVAR= option in the

INFILE statement?

3. What does the INTNX function do?

185

Chapter Review
4. What is the difference between a SAS data file and a
SAS data view?

187 continued...
10-120 Chapter 10 Programmer Efficiency

Chapter Review
5. Why would you use the FILE statement?

6. What is the purpose of the PUT statement?

189
 10.7 Solutions 10-121

10.7 Solutions

Solutions to Exercises
1. Using the FILENAME Statement
a. Open the program p310e01.
b. Use the FILENAME statement to concatenate the three raw data files.
c. Modify the DATA step to use the fileref created in part b to create the SAS data set all_levels.
d. Print the all_levels data set.
p310s01
filename levels ('level_1.dat' 'level_2.dat' 'level_3.dat');

data all_levels;
length Customer_Name $ 40 Customer_Age_Group $ 12
Customer_Type $ 40 Customer_Group $ 40;
infile levels dlm=',';
input Customer_Name $ Customer_Age_Group $ Customer_Type $
Customer_Group $;
run;

proc print data=all_levels;

run;
2. Using the FILEVAR= Option to Read from Raw Data
a. Open the program p310e02.
b. Use the FILEVAR= option to concatenate the three raw data files and create the SAS data set
all_levels.
10-122 Chapter 10 Programmer Efficiency

c. Print the all_levels data set.

p310s02
data all_levels;
drop i;
length Customer_Name $ 40 Customer_Age_Group $ 12
Customer_Type $ 40 Customer_Group $ 40;
do i=1 to 3;
NextFile=cats('level_', i, '.dat'); * Windows and UNIX;
infile levels filevar=NextFile dlm=','
end=Last;
do while (Last=0);
input Customer_Name $ Customer_Age_Group $
Customer_Type $
Customer_Group $;
output;
end;
end;
stop;
run;

proc print data=all_levels;

run;
3. Using the FILEVAR= Option to Read Filenames from a SAS Data Set
a. Use the FILEVAR= option to create a SAS data set named All_Months from the raw data files
named in orion.month_file.
b. Print the first 10 observations of the All_Months SAS data set.
p310s03
data all_months;
set orion.month_file;
format Order_Date Delivery_Date date9.;
infile ORD filevar=File_Name dlm=','
end=LastObs;
do while (not LastObs);
input Customer_ID Order_ID Order_Type
Order_Date : date9. Delivery_Date : Date9.;
output;
end;
run;

proc print data=all_months;

run;
 10.7 Solutions 10-123

4. Creating a DATA Step View

a. Create a view named cc_donations.
1) Read only the observations from the data set orion.employee_donations where the value of
the variable Paid_By is Credit Card.

2) Create a variable named Total_Donations as the total of the variables values for Qtr1, Qtr2,
Qtr3, and Qtr4.
3) Create a new variable Donation_Category with the following values:

Value of Total_Donations Donation_Category

Less than 100 Less than $100

Greater than or equal 100 $100 or more

b. Open and submit the program p310e04 to create a report from the view cc_donations. Use the
variable Donation_Category as a class variable and the variable Total_Donations as an analysis
variable. Verify that the view was created correctly.
p310s04
data cc_donations / view=cc_donations;
set orion.employee_donations;
length Donation_Category $15;
where Paid_By='Credit Card';
Total_Donations=sum(of Qtr1-Qtr4);
if Total_Donations >= 100 then
Donation_Category='$100 or more';
else Donation_Category='Less than $100';
run;

proc means data=cc_donations sum n nonobs maxdec=2;

class Donation_Category;
var Total_Donations;
run;
10-124 Chapter 10 Programmer Efficiency

5. Creating a View and a File in One DATA Step

a. In one DATA step, create a view named younger60 and a file named older60.
b. Read from the data set orion.employee_payroll.
c. Use the variable Birth_Date to calculate each employee's age as of today.
• The view should contain the employees who are younger than 60.
• The file should contain the employees who are 60 or older.
p310s05
data older60 younger60 / view=younger60;
set orion.employee_payroll;
Age=int(yrdif(Birth_Date, today(),'act/act'));
if Age >= 60 then output older60;
else output younger60;
format Birth_Date Employee_Hire_Date Employee_Term_Date
date9.;
run;
d. Attempt to print the file older60 unsuccessfully.
p310s05
proc print data=older60;
title 'Older60 Data Set';
run;
SAS Log
22 proc print data=older60;
ERROR: File WORK.OLDER60.DATA does not exist.
23 title 'Older60 Data Set';
24 run;

NOTE: The SAS System stopped processing this step because of errors.
NOTE: PROCEDURE PRINT used (Total process time):
real time 0.06 seconds
cpu time 0.01 seconds

e. Print the view younger60.

p310s05
proc print data=younger60;
title 'Younger60 Data Set';
run;
f. Print the file older60 successfully.
p310s05
proc print data=older60;
title 'Older60 Data Set';
run;
 10.7 Solutions 10-125

g. Why could you not print older60 in step d?

The file is not created until the view is executed. When you printed the view in part e, the
file was generated.
6. Creating a View with the SQL Procedure and the USING Clause
a. Open the program p310e06 and edit it to assign the libref with the USING clause.
p310s06
proc sql;
create view orion.payroll_donations as
select Employee_ID, Qtr1, Qtr2, Qtr3, Qtr4,
sum(Qtr1, Qtr2, Qtr3, Qtr4) as Total_Donations
from orion.employee_donations
where Paid_By='Payroll Deduction'
using libname orion 's:\Workshop';
quit;
b. Submit a LIBNAME statement to assign a libref of sasdata to the library specified in the table
above.
p310s06
libname sasdata '.'; *Windows/UNIX;
*libname sasdata '.prg3.sasdata'; *z/OS;
*libname sasdata 's:\workshop'; *Windows;
c. Submit a PROC PRINT step to print the view sasdata.payroll_donations.
p310s06
proc print data=sasdata.payroll_donations;
run;
d. Submit a LIBNAME statement to clear the sasdata libref.
p310s06
libname sasdata clear;
10-126 Chapter 10 Programmer Efficiency

7. Creating Two SAS Programs with One DATA Step

a. Write a DATA step to build the two PROC FORMAT steps. Under Windows and UNIX, name the
files Customer_Type.sas and Customer_Group.sas. Under z/OS, name the files
.workshop.sascode(Customer_Type) and .workshop.sascode(Customer_Group).
p310s07
data _null_;
set orion.customer_type end=LastObs;
by Customer_Group_ID;
file 'customer_type.sas'; * Windows and UNIX;
* file '.workshop.sascode(custtype)'; * z/OS;
if _N_=1 then put 'proc format fmtlib; value TypLevl ';
Value=put(Customer_Type_ID, 12.);
Label=put(Customer_Type, $quote50.);
put Value '=' Label ;
if LastObs then put ';run;';
file 'customer_group.sas'; * Windows and UNIX;
* file '.workshop.sascode(cusgroup)'; * z/OS;
if _N_=1 then put 'proc format fmtlib; value GrpLevl ';
Value=put(Customer_Group_ID, 4.);
Label=put(Customer_Group, $quote40.);
if first.Customer_Group_ID then put Value '=' Label ;
if LastObs then put ';run;';
run;
b. Use the %INCLUDE statement to execute the code.
p310s07
/* Windows and UNIX */
%include 'Customer_Type.sas'/source2;
%include 'Customer_Group.sas'/source2;

/* z/OS */
%include '.workshop.sascode(Customer_Type)'/source2;
%include '.workshop.sascode(Customer_Group)'/source2;
 10.7 Solutions 10-127

SAS Log
95 %include 'Customer_Type.sas'/source2;
NOTE: %INCLUDE (level 1) file Customer_Type.sas is file S:\workshop\Customer_Type.sas.
96 +proc format fmtlib;
97 +value TypLevl
98 +1010 ="Orion Club members inactive"
99 +1020 ="Orion Club members low activity"
100 +1030 ="Orion Club members medium activity"
101 +1040 ="Orion Club members high activity"
102 +2010 ="Orion Club Gold members low activity"
103 +2020 ="Orion Club Gold members medium activity"
104 +2030 ="Orion Club Gold members high activity"
105 +3010 ="Internet/Catalog Customers"
106 +;
NOTE: Format TYPLEVL has been output.
106!+ run;

NOTE: PROCEDURE FORMAT used (Total process time):

real time 0.23 seconds
cpu time 0.04 seconds

NOTE: %INCLUDE (level 1) ending.

106
107 %include 'Customer_Group.sas'/source2;
NOTE: %INCLUDE (level 1) file Customer_Group.sas is file S:\workshop\Customer_Group.sas.
108 +proc format fmtlib;
109 +value GrpLevl
110 +10 ="Orion Club members"
111 +20 ="Orion Club Gold members"
112 +30 ="Internet/Catalog Customers"
113 +;
NOTE: Format GRPLEVL has been output.
113!+ run;

NOTE: PROCEDURE FORMAT used (Total process time):

real time 0.00 seconds
cpu time 0.00 seconds

NOTE: %INCLUDE (level 1) ending.

10-128 Chapter 10 Programmer Efficiency

PROC FORMAT Output

„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚ FORMAT NAME: GRPLEVL LENGTH: 26 NUMBER OF VALUES: 3 ‚
‚ MIN LENGTH: 1 MAX LENGTH: 40 DEFAULT LENGTH 26 FUZZ: 0 ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚START ‚END ‚LABEL (VER. V7|V8 18FEB2008:18:08:41)‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚10 ‚10 ‚Orion Club members ‚
‚20 ‚20 ‚Orion Club Gold members ‚
‚30 ‚30 ‚Internet/Catalog Customers ‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒŒ

„ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ†
‚ FORMAT NAME: TYPLEVL LENGTH: 39 NUMBER OF VALUES: 8 ‚
‚ MIN LENGTH: 1 MAX LENGTH: 40 DEFAULT LENGTH 39 FUZZ: 0 ‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ…ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚START ‚END ‚LABEL (VER. V7|V8 18FEB2008:18:08:41)‚
‡ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒˆƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‰
‚1010 ‚1010 ‚Orion Club members inactive ‚
‚1020 ‚1020 ‚Orion Club members low activity ‚
‚1030 ‚1030 ‚Orion Club members medium activity ‚
‚1040 ‚1040 ‚Orion Club members high activity ‚
‚2010 ‚2010 ‚Orion Club Gold members low activity ‚
‚2020 ‚2020 ‚Orion Club Gold members medium activity ‚
‚2030 ‚2030 ‚Orion Club Gold members high activity ‚
‚3010 ‚3010 ‚Internet/Catalog Customers ‚
Šƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ‹ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒŒ
 10.7 Solutions 10-129

8. Sending E-Mail Using the DATA Step

Open the program p310e08, which contains a DATA step with a MERGE statement, and edit the
program to generate e-mails for each employee informing him that the total contribution was mailed.
Under Windows and UNIX, name the file donations.sas. Under z/OS, name the files
.workshop.sascode(donations). Do not include the program file. However, open the program file to
verify that it is correct.
p310s08
proc sort data=orion.employee_addresses out=employee_addresses;
by Employee_ID;
run;
data _null_;
merge orion.employee_donations(in=D) employee_addresses;
by Employee_ID;
if D;
Total_Donation=sum(of Qtr1-Qtr4);
Email=compress(scan(Employee_Name,2,',')||'.'
||scan(Employee_Name,1,',')||'@orion.com');
file 'donations.sas';
put "filename mail email '" Email "' subject='Your Donation';";
put 'data _null_;';
put 'file mail;';
put "put 'Your donation of " Total_Donation dollar4.
" has been sent to " Recipients "';";
put 'run;';
run;
Open the program file, donations.sas, to verify that it is correct.
Partial Contents of donation.sas
filename mail email '[email protected] ' subject='Your Donation';
data _null_;
file mail;
put 'Your donation of $25 has been sent to Mitleid International
90%, Save the Baby Animals 10% ';
run;
filename mail email '[email protected] ' subject='Your Donation';
data _null_;
file mail;
put 'Your donation of $60 has been sent to Disaster Assist, Inc.
80%, Cancer Cures, Inc. 20% ';
run;
10-130 Chapter 10 Programmer Efficiency

9. Writing PROC PRINT Steps Using SQL Dictionary Table Views

a. Use the DATA step to read from Sashelp.VCOLUMN and create a program containing PROC
PRINT steps that create reports for every data set in the orion library that contains the variable
Product_ID. PROC PRINT should only print five observations from each of the data sets.
b. Store the program in a file named Print_Products.sas under Windows and UNIX and
.workshop.sascode(Print_Products) under z/OS.
p310s09
data _null_;
set sashelp.vcolumn;
where Libname='ORION' and Name='Product_ID';
file 'print_products.sas'; * Windows and UNIX;
/* file '.workshop.sascode(prinprod)'; * z/OS; */
Prt=cats(Libname, '.', Memname);
Pgm_Line=catt('proc print data=', Prt, '(obs=5);');
Title=catx(' ', "title 'First Five Observations of", Prt,
"';");
put Pgm_Line;
put Title;
put 'run;';
run;
c. Use the %INCLUDE statement to execute the code or open the program to verify that it is correct.
p310s09
/* Windows and UNIX */
%include 'Print_Products.sas'/source2;

/* z/OS */
%include '.workshop.sascode(Print_Products)'/source2;
10. Using the FCMP Procedure to Store a Formula in a Function
a. Open the program p310e10 and submit it.
b. Use PROC FCMP to encapsulate the IF/THEN logic into a function named KB. Store the
function in work.functions.Marketing.
p310s10
proc fcmp outlib=work.functions.Marketing;
function KB(Quantity, Price);
if Quantity > 2 then return(Quantity * Price / 5);
else return(Quantity * Price / 10);
endsub;
run;
quit;
 10.7 Solutions 10-131

c. Write a DATA step to create a data set named kick_backs that uses the KB function to create a
variable named Kick_Back_Amt. The DATA step should not contain any IF/THEN logic.
p310s10
options cmplib=work.functions;

data kick_backs;
set orion.order_fact(keep=Employee_ID Quantity
Total_Retail_Price);
Kick_Back_Amt=KB(Quantity, Total_Retail_Price);
run;
d. Print the first five observations of the SAS data set kick_backs.
p310s10
proc print data=kick_backs (obs=5);
run;
11. Using the FCMP Procedure to Store a Date Calculation in a Function
a. Open the program p310e11 that contains the formula and a DATA step that generates test data.
b. Write a PROC FCMP step that creates a function named AGE that contains the formula. Store the
function in work.functions.Marketing.
p310s11
proc fcmp outlib=work.functions.Marketing;
function AGE(BirthDate, ActualDate);
return(intck('year', BirthDate, ActualDate)
-(put(BirthDate, mmddyy4.) gt put(ActualDate, mmddyy4.))
+(put(BirthDate, mmddyy4.)||put(ActualDate, mmddyy4.)||
put(ActualDate + 1, mmddyy4.)='022902280301')
);
endsub;
run;
quit;
10-132 Chapter 10 Programmer Efficiency

c. Add an assignment statement to the DATA step in p310e11 that creates a variable named
Real_Age using the AGE function and a variable named Age by using the INTCK function.
p310s11
options cmplib=work.functions;

data real_ages;
do Birth_Date='28feb1960'd to '01mar1960'd;
do Actual_Date='28feb2004'd to '01mar2004'd,
'28feb2005'd to '01mar2005'd;
Real_Age=AGE (Birth_Date, Actual_Date);
Age=intck('year', Birth_Date, Actual_Date);
output;
end;
end;
format Birth_Date Actual_Date worddate.;
run;
d. Print the data to ensure that the function is correctly calculating Real_Age.
p310s11
proc print data=real_ages;
var Birth_Date Actual_Date Real_Age Age;
title1 'Age Calculations using INTCK';
run;
e. Create a data set named customer_ages from the orion.customer_dim data set. The new data set
should contain a new variable named Real_Age using the AGE function with
Customer_BirthDate as the Birth_Date variable and 01JAN2008 as the value of Actual_Date.
The new data set should also contain a new variable named Age calculated using the INTCK
function. Print the first five observations of customer_ages.
p310s11
data customer_ages;
set orion.customer_dim(keep=Customer_ID Customer_Name
Customer_Group Customer_BirthDate);
Real_Age=AGE(Customer_BirthDate,'01jan2008'd);
Age=intck('year',Customer_BirthDate,'01jan2008'd);
format Customer_BirthDate worddate.;
run;

proc print data=customer_ages(obs=5);

title1 'Age Calculations using INTCK;
run;
12. Creating a Function Using DATA SAS File I/O Functions
a. Use PROC FCMP to create a function named NUMS that takes one argument, which is the name
of a SAS data set, and returns the number of logical observations in the data set. Refer to SAS
OnlineDoc or the SAS Help facility to determine which SAS file I/O functions are required.
SAS OnlineDoc Ö Base SAS Ö SAS 9.2 Language Reference Dictionary Ö
Dictionary of Language Elements Ö Functions and CALL Routines by Category
 10.7 Solutions 10-133

b. Open the program p310e12 and submit it. Look in the log to ensure that the function is working
correctly.
p310s12
proc fcmp outlib=work.functions.Marketing;
function NUMS(DSN $);
length DSN $41;
DSID=open(DSN);
return(attrn(DSID, "NLOBSF"));
DSID=close(DSID);
endsub;
run;
quit;

options cmplib=work.functions;

data _null_;
X=NUMS('orion.internet');
put X=;
run;
10-134 Chapter 10 Programmer Efficiency

Solutions to Student Activities (Polls/Quizzes)

10.02 Quiz – Correct Answer

In addition to saving programmer time, does creating a
macro variable or a macro definition always save
computer resources?
No
Why or why not?
An inefficient program is not made more efficient if
you use macro variables or change the program into
a macro definition.

13

10.03 Quiz – Correct Answer

1. Open and submit the program p310a01.
2. How many observations are in the data set quarter?
20
3. Change the files referenced by MON to the current
month and the two previous months. How many
observations are in the data set quarter?
The answer depends on the current month.

You want the program to reference the current

month and the two months without editing the
program every time that it is submitted.

28
 10.7 Solutions 10-135

10.04 Quiz – Correct Answers

If the value of the variable i is the number of the month,
which of the following could be used to create the name
of the raw data file?
a. NextFile=cats("mon",i,".dat");

b. NextFile="mon"||put(i,2.)||".dat";
NextFile=compress(NextFile);

c. NextFile=compress("mon"||put(i,2.)||".dat");

34

10.05 Poll – Correct Answer

Is the STOP statement necessary?
€ Yes
€ No

40
10-136 Chapter 10 Programmer Efficiency

10.06 Multiple Choice Poll – Correct Answer

How many observations are in movingq?
a. One observation per record in all of the raw data files
b. 0
c. 1
d. 3

42

10.07 Poll – Correct Answer

Will the SAS code in p310d08 produce the correct results
if the current month is January or February?
€ Yes
€ No

48
 10.7 Solutions 10-137

10.08 Quiz – Correct Answer

p310d09 contains the following code.
MonNum=month(today());
MidMon=month(intnx('month', today(), -1));
LastMon=month(intnx('month', today(), -2));
Why is the following program more efficient?
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));

The second program creates the Today variable using

the TODAY() function one time, and the assignment
statements reference the variable, Today.

54

10.09 Quiz – Correct Answer

Open and submit the program p310a02.
What does the log report?
The code that is stored in the view
NOTE: DATA step view ORION.MOVINGQ is defined as:

data orion.movingq / view=orion.movingq;

drop MonNum MidMon LastMon I;
retain Today MonNum MidMon LastMon;
if _N_=1 then
do;
Today=today();
MonNum=month(Today);
MidMon=month(intnx('month', Today, -1));
LastMon=month(intnx('month', Today, -2));
end;
do I=MonNum, MidMon, LastMon;
NextFile=cats("mon", i, ".dat");
infile ORD filevar=NextFile dlm=',' end=LastObs;
do while (not LastObs);
input Customer_ID Order_ID Order_Type Order_Date : date9. Delivery_Date : date9.;
output;
end;
end;
stop;
run;

98
10-138 Chapter 10 Programmer Efficiency

10.10 Quiz – Correct Answer

What is the advantage of the following program?
data bonus_view(keep=Manager_ID YrEndBonus)
/ view=bonus_view;
set orion.staff;
YrEndBonus=Salary * 0.05;
where Job_Title contains 'Manager';
run;

proc means data=bonus_view mean sum;

class Manager_ID;
var YrEndBonus;
run;
The program avoids the output associated with the
DATA step.
105 p310d15

10.11 Multiple Choice Poll – Correct Answer

How many times is the rawdata file read?
a. 4
b. 3
c. 2
d. 1
e. 0

115
 10.7 Solutions 10-139

10.12 Multiple Answer Poll – Correct Answers

Which resources can you conserve by using the _NULL_
keyword in the DATA statement?
a. I/O
b. CPU
c. Memory
d. Programmer time

140

10.13 Quiz – Correct Answer

Specify the argument of the MKT function that
corresponds to the each of following variables:
Variable Argument
Customer_ID ID
Delivery_Date Date
Order_Type Type
options cmplib=orion.functions;
data temp;
set orion.order_fact;
Marketing_Comment =
MKT(Customer_ID,Delivery_Date,Order_Type);
run;
169
10-140 Chapter 10 Programmer Efficiency

Solutions to Chapter Review

Chapter Review – Correct Answers

1. What are two parts of the macro facility?
Macro variables and macro definitions

2. What is the purpose of the FILEVAR= option in the

INFILE statement?
The FILEVAR= option names a variable whose
change in value causes the INFILE statement to
close the current input file and open a new one.

3. What does the INTNX function do?

The INTNX function increments a SAS date value
by a given interval or intervals, and returns a SAS
date value.
186 continued...

Chapter Review – Correct Answers

4. What is the difference between a SAS data file and a
SAS data view?
A SAS data file stores the data on disk, whereas a
SAS data view stores the instructions for data
manipulation on disc. For a view, the DATA step or
PROC SQL step is not executed until the view is
referenced.

188 continued...
 10.7 Solutions 10-141

Chapter Review – Correct Answers

5. Why would you use the FILE statement?
Use the FILE statement to specify the external file
in which to write the text generated by the PUT
statements in the current DATA step.

6. What is the purpose of the PUT statement?

The PUT statement writes text to an external
location.

190
10-142 Chapter 10 Programmer Efficiency
Chapter 11 Customizing Your SAS
Session (Self-Study)

11.1 Introduction................................................................................................................... 11-3

11.2 Editing the Configuration File ..................................................................................... 11-7

11.3 Creating an Autoexec.sas File................................................................................... 11-22

11.4 Using the SAS Registry ............................................................................................. 11-28

11.5 Solutions ..................................................................................................................... 11-40

Solutions to Student Activities (Polls/Quizzes) ................................................................... 11-40
11-2 Chapter 11 Customizing Your SAS Session (Self-Study)
 11.1 Introduction 11-3

11.1 Introduction

Objectives
„ Review the OPTIONS procedure.
„ List reasons for customizing a SAS session.
„ Describe the methods that are used to customize a
SAS session.

Using the OPTIONS Procedure (Review)

The OPTIONS procedure lists the current settings
of SAS system options in the SAS log.
General form of the OPTIONS procedure:

PROC OPTIONS <option(s)>;

Option Tasks

DEFINE displays the option's description, type, and group.

VALUE displays the option's value and scope.

LISTGROUPS displays groups and group descriptions (SAS 9.2).

GROUP= displays options belonging to one or more groups.

OPTION= displays a single option.

4
11-4 Chapter 11 Customizing Your SAS Session (Self-Study)

11.01 Quiz
Open and submit the program p311a01.
proc options listgroups;
run;

1. What group would you use to display options used for

procedure output?

2. Change the LISTGROUPS option in the PROC

OPTIONS statement to the GROUP= option to display
the options used for the procedure output that you
identified in part 1.
3. What is the value of the LINESIZE= option?

In SAS 9.2 you can view multiple groups using the following syntax:

PROC OPTIONS GROUP=(group1 group2);

RUN;

p311a02.sas
proc options group=(sort memory);
run;

Reasons for Customizing a SAS Session

You can customize actions taken by SAS when SAS
starts. Aspects of your SAS session that you might want
to control when the session starts include the following:
„ specifying system options in effect when the session
starts
„ executing SAS statements such as LIBNAME
statements automatically
„ changing the working and default folders for Open
and Save As actions when you use the SAS
windowing environment

8
 11.1 Introduction 11-5

Techniques for Customizing a SAS Session

These three techniques can be used to customize
a SAS session:
„ SAS configuration file

„ autoexec file

„ SAS Registry

Defining the Three Techniques

The three techniques are defined in the table below.
configuration file an external file that contains SAS
system options used to establish
the SAS session
autoexec file a file that contains SAS statements
that are executed immediately after
SAS initializes and before any user
input is accepted
SAS Registry a type of SAS file called an item
store that contains configuration
data for one or more SAS software
products

10
11-6 Chapter 11 Customizing Your SAS Session (Self-Study)

Comparing the Three Techniques

Configuration File Autoexec File SAS Registry
Easy to change Easiest to change Change with caution
a text file named a SAS program with a a SAS file named
sasv9.cfg in Windows .sas extension in regstry.sas7bitm
and UNIX Windows and UNIX

processed before SAS processed after SAS processed during

initializes initializes initialization
can be edited with a SAS can be edited with any can be changed with
text editor or an ASCII text editor PROC REGISTRY code
text editor or by using the
REGISTRY window

contains only system contains SAS code such consists of keys and
options and the location as OPTIONS statements sub-keys that refer to
of SAS components or LIBNAME statements particular aspects of SAS

11

z/OS users can create a user configuration file using any text editor to write SAS system options into a
physical file. The physical file can then be specified in the CONFIG= invocation system option
interactively or in batch mode.

11.02 Multiple Answer Poll

Which of the following have you investigated?
a. SAS configuration file
b. autoexec.sas file
c. SAS Registry
d. None of the above

13
 11.2 Editing the Configuration File 11-7

11.2 Editing the Configuration File

Objectives
„ Define the purpose of the configuration file.
„ List the two parts of the configuration file.
„ Create a custom configuration file.
„ Use the custom configuration file.

16

Defining the Configuration File

The SAS configuration file is a text file that contains
SAS system options and SAS installation locations
used to establish the SAS session.
„ In Windows and UNIX, the configuration file has
an extension of .cfg.
„ You can view the configuration file using any text
editor, including the SAS Program Editor window.

17
11-8 Chapter 11 Customizing Your SAS Session (Self-Study)

Purpose of the Configuration File

Three types of system options are in the configuration file:
„ initialization options (for example, –SGIO)

„ options that can be specified in an OPTIONS

statement (for example, FULLSTIMER)
„ the –SET option that creates operating system
environment variables that indicate the location of the
following files:
– SAS Help and Documentation files
– message files
– pathnames to the SAS executable files

18

Importance of the Configuration File

„ You must have at least one configuration file in order
for SAS to initialize.
„ You can have multiple configuration files that are all
processed when your SAS session begins.

The SAS configuration file is particularly important

because it specifies the folders that are searched
for the various components of SAS products.

19
 11.2 Editing the Configuration File 11-9

Viewing the Configuration File

The configuration file is divided into two sections
separated by a warning note.
„ The first section specifies SAS system options that
are not updated by the SAS setup applications.
Put system options in this section.
„ The second section is used by the SAS setup
applications for updating information about where
SAS software is installed. Do not edit below this line.

20
11-10 Chapter 11 Customizing Your SAS Session (Self-Study)

The Top of the Configuration File

/* set default locations */
-fontsloc "!sasroot\core\resource"
This is the
-TRAINLOC "" section that you can edit.
/* set the default fileref for the PARMCARDS= option */
Only system options can be added here.
-SET FT15F001 'FT15F001.DAT'
Those system options are in effect when SAS initializes.
/*---------------------------------------------------------------\
| SAS System FORMCHARS, used by pressing ALT then the decimal |
| number for the Extended ASCII character. |
\---------------------------------------------------------------*/
/* This is the OEM character set */
/* -FORMCHAR "³ÄÚ¿ÃÅ ÀÁÙ+=|-/\<>*" */
/* This is the ANSI character set (for SAS Monospace font and
ANSI Sasfont) */
-FORMCHAR "‚ƒ„…†‡ˆ‰Š‹Œ+=|-/\<>*"
/* This is the ANSI character set */
/* -FORMCHAR "|----|+|---+=|-/\<>*" */

This is the section that you can edit. Only system

options can be added here. Those system options
are in effect when SAS initializes.
21

System Option Purpose

-FONTSLOC specifies the location that contains the SAS fonts that are loaded by some
Universal Printer drivers.

-TRAINLOC specifies the base location of SAS online training courses.

-SET defines a SAS (internal) environment variable. In this case, the variable,
FT15F001, specifies the file reference of a file that SAS opens when it
encounters a PARMCARDS (or PARMCARDS4) statement in a
procedure. The PARMCARDS statement is used in the BMDP and
EXPLODE procedures.

-FORMCHAR specifies formatting characters used to construct tabular output outlines

and dividers for various procedures, such as the FREQ and TABULATE
procedures. If you omit formatting characters as an option in the
procedure, the default specifications given in the FORMCHAR= system
option are used. Notice that you can also specify a hexadecimal character
constant as a formatting character. When you use a hexadecimal constant
with this option, SAS interprets the value of the hexadecimal constant as
appropriate for your operating system.
 11.2 Editing the Configuration File 11-11

The Warning Message in the Configuration File

/*---------------------------------------------------------------\
| WARNING: INSTALL Application edits below this line. User |
| options should be added above this box comment. |
| INSTALL Application maintains and modifies the |
| following options; -SASAUTOS, -SASHELP, -SASMSG, |
| -PATH, and -MAPS. It also maintains and modifies |
| the following CONFIG variables with the -SET option; |
| INSTALL, USAGE, LIBRARY, SAMPSIO, SAMPSRC, SASCBT, |
| and SASEXT01-SASEXT50. It preserves all lines above |
| the line containing 'DO NOT EDIT BELOW THIS LINE'. |
\---------------------------------------------------------------*/
/* DO NOT EDIT BELOW THIS LINE - INSTALL Application edits below this line */
/* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * */

22
11-12 Chapter 11 Customizing Your SAS Session (Self-Study)

Partial Configuration File Setup Information

-SET sasext0 "C:\Program Files\SAS\SASFoundation\9.2"
-SET sasroot "C:\Program Files\SAS\SASFoundation\9.2"
-SET sasext1 "C:\Program Files\SAS\SASFoundation\9.2\nls"
/* Setup the MYSASFILES system variable */
-SET MYSASFILES "?CSIDL_PERSONAL\My SAS Files\9.2"
/* Setup the default SAS System user profile folder */
-SASUSER "?CSIDL_PERSONAL\My SAS Files\9.2"
/* Setup the default SAS System user work folder */
-WORK "!TEMP\SAS Temporary Files"
/* Setup the SAS System configuration folder */
-SET SASCFG "C:\Program Files\SAS\SASFoundation\9.2\nls\en"
/* location of help in OS help format */
-HELPLOC ("!MYSASFILES\classdoc" "!sasroot\nls\en\help"
"!sasroot\core\help")

<lines removed>

This is the section that you should not edit. This

provides the information about where SAS is
installed.
23

The following are SAS librefs, which are established in the configuration file:

SAS System Libref What the Option Does

Sasuser points to the location of the Sasuser library.

Work points to the high-level location of the Work library.

HELPLOC points to and concatenates the locations of the SAS Help facility
files.

The following are SAS internal environment variables, which are set in the configuration file:

Environment What the Variable Does

Variable

sasroot sets the location where SAS software was installed, often referred to
as the current or working location or only as sasroot.

sasext0 sets the location for additional SAS software modules.

sasext1 sets the location for the National Language Support modules.

MYSASFILES sets the location for the Sasuser libref as an environment variable.

sascfg sets the location for the .sasv9 configuration file.

 11.2 Editing the Configuration File 11-13

Configuration Files in Windows and UNIX

The following information applies to Windows and UNIX:
„ The configuration file typically resides in the directory
where SAS was installed. By default, this directory is
the !SASROOT directory.
„ The systems administrator can edit this configuration
file so that it contains appropriate options for your site.
„ You can create a customized configuration file.

In UNIX, there can be a restricted configuration

file containing options that you cannot change.

24

Reference Information

Location of the Configuration File in the Windows Operating Environment

In previous releases of SAS, the default configuration file was stored in the !SASROOT folder. The
!SASROOT folder is the folder in which you install SAS.
Starting with SAS®9, SAS creates two default configuration files during installation. Both configuration
files are named SASV9.CFG.
SAS stores one of these files in the !SASROOT folder and the other in the !sasroot\nls\language-code
folder. The language-code is a two-letter language code that indicates the SAS default language.
The SASV9.CFG file that is located in the !SASROOT folder contains a CONFIG system option that
specifies the location of the configuration file for the SAS default language. The default system options
that are used to start SAS are specified in the !sasroot\nls\language-code\SASV9.CFG file. For example,
if SAS is installed in the default folder and the default language is English, the SASV9.CFG file in the
!SASROOT folder contains this line:
-config "c:\program files\SAS\SASFoundation\9.2\nls\en\sasv9.cfg"
Use the RESTRICT option in the PROC OPTIONS statement to view the restricted options.

PROC OPTIONS RESTRICT;

RUN;
11-14 Chapter 11 Customizing Your SAS Session (Self-Study)

Configuration Files in z/OS

In z/OS, SAS uses two types of configuration files:
„ the system configuration file, which is used by all
users at your site by default. Your on-site SAS support
personnel maintain the system configuration file for
your site.
„ a user configuration file, which is generally used
by an individual user or department.

25

Creating a Custom Configuration File

These steps enable creating a custom configuration file:
1. Use a text editor to write SAS system options into
a file.
2. Specify one or more system options, and apply the
syntax that you use when you type those options
at an operating system command prompt.
3. Save the file.

 Specifics for each operating environment

are described on subsequent slides.

26
 11.2 Editing the Configuration File 11-15

UNIX Specifics: Creating a Configuration File

Example:

-nocenter
-nodate
-msglevel i
-linesize 64
-pagesize 56
-work /users/myuserid/tmp

Save the file to either a sasv9.cfg or .sasv9.cfg file.

27

UNIX Specifics: –WORK System Option

The –WORK SAS system option specifies the location
of the Work library.

-WORK pathname

pathname specifies the directory (not a filename)

where your Work SAS library can be
created or found. SAS will create the
directory if it does not exist. You must
have Write permission to the location
specified in pathname.

28
11-16 Chapter 11 Customizing Your SAS Session (Self-Study)

Windows Specifics:
Creating a Configuration File
To ensure that all of the required system options are
defined in the custom configuration file, copy the default
file and modify the copy.
„ Name the file sasv9.cfg or .sasv9.cfg.

„ Store the file in the Windows user-profile folder or use

the –CONFIG option when you invoke SAS.

 In Windows XP, the path for the Windows user-profile

folder is as follows:
c:\Documents and Settings\user-id\
My Documents\My SAS Files\9.2

29 continued...

Windows Specifics:
Creating a Configuration File
Example:
-nocenter
-nodate
-msglevel i
-linesize 64
-pagesize 56
-work "c:\temp"
-sasinitialfolder s:\workshop

/* set default locations */

-fontsloc "!sasroot\core\resource"
-TRAINLOC ""
/* set the default fileref for the PARMCARDS= option */
-SET FT15F001 'FT15F001.DAT'
/*---------------------------------------------------------------\
| SAS System FORMCHARS, used by pressing ALT then the decimal |
| number for the Extended ASCII character. |
\---------------------------------------------------------------*/
/* This is the OEM character set */
/* -FORMCHAR "³ÄÚ¿ÃÅ ÀÁÙ+=|-/\<>*" */
/* This is the ANSI character set (for SAS Monospace font and ANSI Sasfont) */
-FORMCHAR "‚ƒ„…†‡ˆ‰Š‹Œ+=|-/\<>*"
/* This is the ANSI character set */
/* -FORMCHAR "|----|+|---+=|-/\<>*" */

30
 11.2 Editing the Configuration File 11-17

Windows Specifics: –WORK System Option

The –WORK SAS system option specifies the location
of the Work library.

-WORK "library-specification"

"library- specifies an environment variable or a

specification" Windows pathname. The value of
library-specification must resolve to a
valid Windows directory or
subdirectory pathname. The library-
specification value must be enclosed
in double quotation marks.

31

Windows Specifics: –SASINITIALFOLDER

In the Windows operating environment, the
-SASINITIALFOLDER SAS system option changes the
working folder and the default folders for the Open and
Save As dialog boxes to the specified folder after SAS
initialization is complete.

-SASINITIALFOLDER newfolder

newfolder specifies the path to the current working

folder and the default folders for the Open
and Save As dialog boxes. If newfolder
contains spaces, it must be enclosed in
quotation marks.

32
11-18 Chapter 11 Customizing Your SAS Session (Self-Study)

z/OS Specifics: Creating a Configuration File

Use any text editor to write SAS system options into
a physical file.
„ The configuration file can be either a sequential data
set or a member of a partitioned data set and can
have either fixed length or variable length records.
„ Each line of a configuration file can contain one or
more system options. If you specify more than one
system option on a line, use either a blank space
or a comma to separate the options.

33 continued...

z/OS Specifics: Creating a Configuration File

Example:

nocenter
nodate
msglevel=i
linesize=64
pagesize=56
-work userid.myfile.mywork

When you specify SAS system options in a

configuration file, blank spaces are not permitted
before or after an equal sign.

34
 11.2 Editing the Configuration File 11-19

z/OS Specifics: –WORK System Option

The –WORK SAS system option specifies the location
of the Work library.

-WORK library-specification

library- can be a DDNAME that was previously

specification associated with a SAS library using JCL
or TSO commands or the name of a
physical file that comprises a SAS data
library.

35

11.03 Quiz
In the Windows operating environment, navigate to
C:\Program Files\SAS\SASFoundation\9.2\nls\en\sasv9.cfg

Which option is set to open the SAS windowing

environment and the Explorer window?

Should you edit that option?

37
11-20 Chapter 11 Customizing Your SAS Session (Self-Study)

UNIX: Specifying a Configuration File

One way to specify a configuration file is to use the
-CONFIG system option in the SAS command.

sas -config name-of-configuration-file

„ When you specify the –CONFIG option, SAS still

processes restricted configuration files and uses
the restricted settings instead of those specified
in the custom configuration file.

39

For more information, consult the SAS Help facility by following the path described below:
Using SAS Software in Your Operating Environment Ö
SAS 9.2 Companion for UNIX Environments Ö Running SAS Software Under UNIX Ö
Getting Started with SAS in UNIX Environments Ö
Customizing Your SAS Session by Using Configuration and Autoexec Files

UNIX: Specifying a Configuration File

If you do not use the -CONFIG option, SAS processes the
configuration files in the following order:
1. sasv9.cfg in the !SASROOT directory
2. sasv9_local.cfg in the !SASROOT directory
3. .sasv9.cfg in your home directory
4. sasv9.cfg in your home directory
5. sasv9.cfg in your current directory
6. any restricted configuration files

40
 11.2 Editing the Configuration File 11-21

Windows: Specifying a Configuration File

When you use a file that is located in a different folder or
that has a different name from your default configuration
file, use the –CONFIG system option to specify the
location of the configuration file.
For example, the Target field of the SAS Properties
dialog box for the SAS icon shortcut might contain the
following:

"c:\program files\SAS\SASFoundation\9.2\sas.exe"
-config "c:\mysas\mysasconfig.CFG"
41

For more information, consult the SAS Help facility by following the path described below:
Using SAS Software in Your Operating Environment Ö SAS 9.2 Companion for Windows Ö
Running SAS under Windows Ö Getting Started Ö Files Used by SAS

z/OS: Specifying a Configuration File

To tell SAS where to find your user configuration file,
do the following:
„ If you use the SAS cataloged procedure to invoke
SAS in batch mode, use the CONFIG= parameter,
for example:
//S1 EXEC SAS,CONFIG='MY.CONFIG.FILE'
„If you use the SAS CLIST or SASRX exec to invoke
SAS under TSO, use the CONFIG operand, for
example:
sas config('''my.config.file''') or
sasrx -config 'my.config.file'

42

For more information, consult the SAS Help facility by following the path described below:
Using SAS Software in Your Operating Environment Ö
SAS 9.2 Companion for z/OS Ö Running SAS Software under z/OS Ö
Initializing and Configuring SAS Software Ö Customizing Your SAS Session
11-22 Chapter 11 Customizing Your SAS Session (Self-Study)

11.3 Creating an Autoexec.sas File

Objectives
„ Define an autoexec file.
„ Create an autoexec file.
„ Execute the autoexec file.

45

Defining an Autoexec File

An autoexec file has the following characteristics:
„ It is a SAS program file with a file extension of .sas in
Windows and UNIX.
„ It contains SAS statements that are executed
immediately after SAS initializes and before any user
input is accepted. These SAS statements can be used
to invoke SAS programs automatically, set up librefs
and/or filerefs for use during your SAS session, or set
system options.
„ It is not required in order to run SAS.

46
 11.3 Creating an Autoexec.sas File 11-23

11.04 Poll
Have you ever created an autoexec file?
€ Yes
€ No

48

Example of an Autoexec File

libname orion '/user/workshop';
options fmtsearch=(orion.myfmts orion)
UNIX: nodate nonumber ls=80;
%include 'p303d01.sas';

libname orion 's:\workshop';

options fmtsearch=(orion.myfmts orion)
Windows: nodate nonumber ls=80;
%include 'p303d01.sas';

libname orion '.prg3.sasdata';

options fmtsearch=(orion.myfmts orion)
z/OS: nodate nonumber ls=80;
%include '.prg3.sascode(p303d01)';
p311d01
49
11-24 Chapter 11 Customizing Your SAS Session (Self-Study)

Using an Autoexec File

„ There is no autoexec file by default.
„ To create one, use a text editor to create a program
and save the program in the appropriate location.
„ If you do have an autoexec file, the default name
is autoexec.sas.

The appropriate location is dependent on your

operating environment.

50

UNIX: Location for the Autoexec.sas File

In the UNIX operating environment, you can save the
autoexec file in any folder. SAS uses the following search
order to find the autoexec.sas file:
1. your current directory
2. your home directory
3. the !SASROOT directory

51
 11.3 Creating an Autoexec.sas File 11-25

Windows: Location for the Autoexec.sas File

In the Windows operating environment, you can save the
autoexec file in any folder. SAS uses the following search
order to find the autoexec.sas file:
1. the current folder
2. the paths that are specified by the Windows PATH
environment variable
3. the root folder of the current drive
4. the folder that contains the sas.exe file

52

z/OS: Location for the Autoexec File

„ Under z/OS, an autoexec file can be either a
sequential data set or a member of a partitioned data
set.
„ You must specify the location of the autoexec file
when you invoke SAS.
Method of Syntax for Using the AUTOEXEC File
Invoking SAS
Command line sas autoexec('''my.auto.exec''')
invocation
Batch //MYJOB EXEC SAS
//SASEXEC DD DSN=MY.AUTO.EXEC,DISP=SHR

53
11-26 Chapter 11 Customizing Your SAS Session (Self-Study)

11.05 Poll
Is the code from the autoexec file included as part of your
log?
€ Yes
€ No

55

Using the ECHOAUTO System Option

The configuration option, ECHOAUTO, determines
whether the code in the autoexec file is written to the log.

NOECHOAUTO | ECHOAUTO

NOECHOAUTO specifies that SAS source lines that are read

from the autoexec file are not printed in the SAS
log, even though they are executed (DEFAULT).
ECHOAUTO specifies that SAS source lines that are read
from the autoexec file be printed in the SAS log.

Regardless of the setting of this option, messages

that result from errors in the autoexec files are
printed in the SAS log.
57
 11.3 Creating an Autoexec.sas File 11-27

Disabling the Autoexec File

You can use the NOAUTOEXEC SAS system option
at invocation to specify that SAS is not to process any
autoexec files.
Windows/UNIX

sas -noautoexec
z/OS

sas autoexec(noautoexec)

58
11-28 Chapter 11 Customizing Your SAS Session (Self-Study)

11.4 Using the SAS Registry

Objectives
„ Define the SAS Registry.
„ Investigate techniques for modifying the SAS Registry.

61

Defining the SAS Registry

The SAS Registry is the central storage area for
configuration data for SAS. Customizations to the SAS
Registry remain in effect for more than one SAS session.
Examples of what the registry stores are listed below:
„ the libraries and file shortcuts that SAS assigns
at startup
„ the menu definitions for Explorer pop-up menus

„ the printers that are defined for use

„ configuration data for various SAS products

62
 11.4 Using the SAS Registry 11-29

Storage Location for the SAS Registry

The registry consists of two parts:
„ One part is stored in the Sashelp library.

„ The other part is stored in the Sasuser library.

The SAS Registry is not displayed in the

SAS Explorer window.

63

Techniques for Modifying the Configuration

Changes to configuration are saved in the SAS Registry
when you make changes using tools such as the
following:
„ the New Library window

„ the Universal Print windows

„ the Explorer Options window

64
11-30 Chapter 11 Customizing Your SAS Session (Self-Study)

Using the New Library Window

65

Using the Print Setup Window

Universal printers should be configured by using either the
PRTDEF procedure or the Print Setup window.

66
 To open the Print Setup window, select File Ö Print Setup.
 11.4 Using the SAS Registry 11-31

The PRTDEF procedure creates printer definitions in batch mode either for an individual user or for all
SAS users at your site. Your system administrator can create printer definitions in the SAS Registry and
make these printers available to all SAS users at your site by using PROC PRTDEF with the
USESASHELP option. An individual user can create personal printer definitions in the SAS Registry by
using PROC PRTDEF.

PROC PRTDEF <option(s)>;

Option Task

DATA= specifies the input data set that contains the printer
attributes.

DELETE specifies that the default operation is to delete the

printer definitions from the registry.

FOREIGN specifies that the registry entries are created for export
to a different host.

LIST specifies that a list of printers that are created or

replaced will be written to the log.

REPLACE specifies that any printer name that already exists will
be modified by using the information in the printer
attributes data set.

USESASHELP specifies whether the printer definitions are available

to all users or only the users running PROC PRTDEF.

Customizing the SAS Explorer

Use the Explorer Options window to configure Explorer
settings.

 To open the Explorer Options window, with the Explorer

67 window active, select Tools Ö Options Ö Explorer.
11-32 Chapter 11 Customizing Your SAS Session (Self-Study)

Techniques for Modifying the Configuration

You can also make changes to the configuration by using
one
of the following techniques:
„ the interactive SAS Registry Editor

„ the REGISTRY procedure

The SAS Registry is designed for use by system

administrators and experienced SAS users.

Do not make a mistake when you edit the registry.

Your SAS system might become unstable or
unusable. This can negatively affect any SAS session.

68

11.06 Quiz
Open the Registry Editor by selecting
Solutions Ö Accessories Ö Registry Editor
or use the REGEDIT command on the command line.
Which key would contain settings from the LIBNAME
window?

70
 11.4 Using the SAS Registry 11-33

Using the Registry Editor

 To open the Registry Editor window, select

Solutions Ö Accessories Ö Registry Editor or
72
use the REGEDIT command on the command line.

Viewing the LIBNAME Keys

To remove this information

from the registry safely,
right-click on the libref
in the SAS Explorer window
and select Delete.

73
11-34 Chapter 11 Customizing Your SAS Session (Self-Study)

Viewing Printer Keys

74
 11.4 Using the SAS Registry 11-35

Viewing the SAS Explorer Keys

Use the Registry Editor to view the current Explorer
settings in the SAS Registry.

75

Registry Key What Portion of the Explorer It Configures

CORE\EXPLORER\CONFIGURATION The portions of the Explorer that are initialized at startup.

CORE\EXPLORER\MENUS The context menus that are displayed in the Explorer.

CORE\EXPLORER\KEYEVENTS The valid key events for the 3270 interface. This key is
used only on the mainframe platforms.

CORE\EXPLORER\ICONS The icons displayed in the Explorer. If the icon value is -1,
this causes the icon to be hidden in the Explorer.

CORE\EXPLORER\NEW What types of objects are available from the File Ö New
menu in Explorer.
11-36 Chapter 11 Customizing Your SAS Session (Self-Study)

Using the Registry Procedure

The REGISTRY procedure enables you to maintain
the SAS Registry.
You can create registry files with the SAS Registry Editor
or with any text editor.
A registry file must have a particular structure:
„ Each entry in the registry file consists of a key name,
followed on the next line by one or more values.
„ The key name identifies the key or subkey that you
are defining.
„ Any values that follow specify the names or data
to associate with the key.

76

The REGISTRY procedure enables you to do the following:

• import registry files to populate the Sashelp and Sasuser registries
• export all or part of the registry to another file
• list the contents of the registry in the SAS log
• compare the contents of the registry to a file
• uninstall a registry file
• deliver detailed status information about when a key or value will be overwritten or uninstalled
• delete entries in the Sasuser registry
• validate that the registry exists
• list diagnostic information
 11.4 Using the SAS Registry 11-37

Using the REGISTRY Procedure

General form of the REGISTRY procedure:

PROC REGISTRY <option(s)>;

RUN;

Selected Options Use

COMPAREREG1= and compare two registry files.
COMPAREREG2=
COMPARETO= compares the contents of a registry to a
file.
EXPORT= writes the contents of a registry to the
specified file.
LISTHELP writes the contents of the Sashelp portion
of the registry to the SAS log.
77

Using the REGISTRY Procedure Partial Log

201 proc registry listuser;
Example: NOTE: Contents of SASUSER REGISTRY.
[ HKEY_USER_ROOT]
proc registry listuser; [
[
CORE]
EXPLORER]
run; [ CONFIGURATION]
[ CUSTOM LISTVIEWS]
[ LV_XEXPLIN]
[ COLUMN1]
Column Name="Name"
 The LISTUSER option [ COLUMN2]
Column Name="Engine"
reports the contents [ COLUMN3]
Column Name="Type"
of the Sasuser portion [ COLUMN4]
Column Name="Host Pathname"
of the SAS Registry [ COLUMN5]
Column Name="Modified"
in the log. [
[
DMSEXP]
COLUMN1]
Column Width=int:108
[ COLUMN2]
Column Width=int:72
[ COLUMN3]
Column Width=int:108
[ COLUMN4]
Column Width=int:315
[ COLUMN5]
Column Width=int:158

78 p311d02 continued...
11-38 Chapter 11 Customizing Your SAS Session (Self-Study)

Using the REGISTRY Procedure

[ OPTIONS]
[ LIBNAMES]
[ ORION]
ENGINE=" " Key for
LIBRARY="orion "
OPTIONS="" the library
PATH="S:\Workshop" definition
[ PRINTING]
Print File=""
that was
[ REGEDIT] enabled
[ INIT] at startup
colwidth=int:180
treewidth=int:230
[ PRODUCTS]
[ BASE]
[ Enhanced Editor]
[ Use_As_Default]
[ Viewtable]
Height=int:102
HorizontalPosition=int:0
VerticalPosition=int:36
Width=int:100
NOTE: PROCEDURE REGISTRY used (Total process time):
real time 0.00 seconds
cpu time 0.00 seconds
p311d02
79

Using the REGISTRY Procedure

proc registry list
startat='core\printing\paper sizes';
run;

LIST writes the contents of the entire SAS

registry to the SAS log.
STARTAT=key starts exporting or writing or comparing
the contents of a registry at the
specified key.

 The LIST option generates a great deal of output

without the STARTAT= option.

p311d03
80
 11.4 Using the SAS Registry 11-39

Using the REGISTRY Procedure

Partial Log
00 proc registry list startat='core\printing\paper sizes';
401 run;

NOTE: Contents of SASHELP REGISTRY starting at subkey [core\printing\paper sizes]

core\printing\paper sizes]
[ 13x18]
Height=double:18
Units="IN"
Width=double:13
[ 16K]
Height=double:10.75
Units="IN"
Width=double:7.75
[ 24x108]
Height=double:108
Units="IN"
Width=double:24
[ 24x48]
Height=double:48
Units="IN"
Width=double:24

81

11.07 Quiz
Open and submit p311a03.
p311a03
proc registry list startat='core\options\libnames';
run;

Are there any libraries listed?

83
11-40 Chapter 11 Customizing Your SAS Session (Self-Study)

11.5 Solutions

Solutions to Student Activities (Polls/Quizzes)

11.01 Quiz
Open and submit the program p311a01.
proc options listgroups;
run;

1. What group would you use to display options used for

procedure output?

2. Change the LISTGROUPS option in the PROC

OPTIONS statement to the GROUP= option to display
the options used for the procedure output that you
identified in part 1.
3. What is the value of the LINESIZE= option?

proc options group=listcontrol;

run;

11.03 Quiz – Correct Answers

In the Windows operating environment, navigate to
C:\Program Files\SAS\SASFoundation\9.2\nls\en\sasv9.cfg

Which option is set to open the SAS windowing

environment and the Explorer window?
-dmsexp

Should you edit that option?

No

38
 11.5 Solutions 11-41

11.05 Poll – Correct Answer

Is the code from the autoexec file included as part of your
log?
€ Yes
€ No

56

11.06 Quiz – Correct Answer

Open the Registry Editor by selecting
Solutions Ö Accessories Ö Registry Editor
or use the REGEDIT command on the command line.
Which key would contain settings from the LIBNAME
window?
Core

71
11-42 Chapter 11 Customizing Your SAS Session (Self-Study)

11.07 Quiz – Correct Answer

Open and submit p311a03.
p311a03
proc registry list startat='core\options\libnames';
run;

Are there any libraries listed?

No, if you used a LIBNAME statement to assign the
libref orion
Yes, if you used the Enable at Startup check box in
the New Libraries window to assign the libref orion

84
Chapter 12 Learning More

12.1 Conclusions .................................................................................................................. 12-3

12.2 SAS Resources ............................................................................................................. 12-8

12.3 Beyond This Course ................................................................................................... 12-12

12-2 Chapter 12 Learning More
 12.1 Conclusions 12-3

12.1 Conclusions

Objectives
„ Review techniques for conserving computer
resources.

Techniques for Reducing I/O Operations

The techniques for reducing I/O operations include the
following:
„ Minimize the number of variables and observations.

„ Reduce the number of times that the data is

processed.
„ Use a SAS data file to process the same raw data file
repeatedly.
„ Use the SASFILE statement to process a small
SAS data set repeatedly.

4 continued...
12-4 Chapter 12 Learning More

Techniques for Reducing I/O Operations

„ Minimize the size of the SAS data set.
„ Use appropriate BUFSIZE= and/or BUFNO= options
for random or sequential access.
„ Bypass system file caching in Windows and UNIX.
„ Create views in programs that require intermediate
temporary SAS data files.
„ Create indexes on variables used for WHERE
processing.

Techniques for Reducing Data Set Size

The techniques for reducing data set size include the
following:
„ Store integers as reduced-length numerics.

„ Compress the data set.

 Reducing the size of a SAS data set reduces

the I/O required to process it.

6
 12.1 Conclusions 12-5

Reducing Memory Usage

The techniques for reducing memory usage include the
following:
„ Use small data set page sizes when you create data
sets that will be accessed in a sparse, random pattern.
„ Use a single read buffer when the data is accessed
randomly instead of sequentially.
„ Use BY-group processing instead of CLASS
statements in those procedures that support both,
especially where you have pre-sorted data or can use
an existing index.

7
12-6 Chapter 12 Learning More

Selected Additional Resources for Specific Topics

Permanently Store and Use Formats

• http://support.sas.com/faq/018/FAQ01816.html

Functions by Category
• http://support.sas.com/onlinedoc/913/getDoc/en/lrdict.hlp/a000245860.htm

Creating Custom Date and Number Formats

• http://support.sas.com/techsup/unotes/SN/008/008510.html
• http://support.sas.com/onlinedoc/913/getDoc/en/proc.hlp/a002473467.htm

Using the DATA Step Merge or the SQL Procedure

• http://support.sas.com/techsup/technote/ts644.pdf
• http://support.sas.com/techsup/technote/ts705.pdf
• http://support.sas.com/techsup/technote/ts553.html
• http://support.sas.com/techsup/technote/ts320.html
• http://support.sas.com/resources/papers/sgf09/336-2009.pdf
• http://support.sas.com/resources/papers/proceedings09/037-2009.pdf

Pipes and Threads: Performance Testing

• http://www2.sas.com/proceedings/forum2007/196-2007.pdf

SAS Efficiency
• http://www2.sas.com/proceedings/forum2007/042-2007.pdf
• http://www2.sas.com/proceedings/forum2007/209-2007.pdf

Hash Tables
• http://www2.sas.com/proceedings/forum2007/039-2007.pdf
• http://www2.sas.com/proceedings/sugi31/244-31.pdf
• http://support.sas.com/resources/papers/sgf2008/hashing92.pdf
• http://support.sas.com/rnd/base/datastep/dot/hash-getting-started.pdf
• http://support.sas.com/rnd/base/datastep/dot/iterator-getting-started.pdf

Arrays
• http://support.sas.com/rnd/papers/sgf07/arrays1780.pdf

Numeric Precision
• http://support.sas.com/techsup/technote/ts654.pdf

What's New in SAS 9.2

• http://support.sas.com/software/index.html

The FCMP Procedure

• http://www2.sas.com/proceedings/forum2007/008-2007.pdf
 12.1 Conclusions 12-7

Solving SAS Performance Problems: Employing Host-Based Tools

• http://support.sas.com/rnd/papers/sugi31/practicalperf.pdf
• http://support.sas.com/techsup/technote/ts684/ts684.html

Configuring SAS I/O Subsystem

• http://support.sas.com/rnd/papers/sgf07/sgf2007-iosubsystem.pdf
• http://www.sas.com/partners/directory/hp/sasapp.pdf
• http://www2.sas.com/proceedings/forum2007/203-2007.pdf
• www.nesug.info/Proceedings/nesug07/as/as04.pdf
• http://support.sas.com/resources/papers/proceedings09/310-2009.pdf

Threading
• http://www2.sas.com/proceedings/sugi29/217-29.pdf
• http://www2.sas.com/proceedings/sugi28/282-28.pdf

Scalability and Performance Papers

• http://support.sas.com/rnd/scalability/papers/index.html
12-8 Chapter 12 Learning More

12.2 SAS Resources

Objectives
„ Identify areas of support that SAS offers.
„ List additional resources.

Education
Comprehensive training to deliver greater value to your
organization

„ More than 200 course offerings

„ World-class instructors
„ Multiple delivery methods: instructor-led and
self-paced
„ Training centers around the world

http://support.sas.com/training/

10
 12.2 SAS Resources 12-9

SAS Publishing
SAS offers a complete selection of publications to help
customers use SAS software to its fullest potential:

„ Multiple delivery methods: e-books,

CD-ROM, and hard-copy books
„ Wide spectrum of topics
„ Partnerships with outside authors,
other publishers, and distributors

http://support.sas.com/publishing/

11

SAS Global Certification Program

SAS offers several globally recognized certifications.

„ Computer-based
certification exams –
typically 60-70 questions
and 2-3 hours in length
„ Preparation materials and
practice exams available
„ Worldwide directory of
SAS Certified Professionals

http://support.sas.com/certify/

12
12-10 Chapter 12 Learning More

Support
SAS provides a variety of self-help and assisted-help
resources.

„ SAS Knowledge Base

„ Downloads and hot fixes
„ License assistance
„ SAS discussion forums
„ SAS Technical Support

http://support.sas.com/techsup/

13

User Groups
SAS supports many local, regional, international, and
special-interest SAS user groups.
„ SAS Global Forum

„ Online SAS Community: www.sasCommunity.org

http://support.sas.com/usergroups/

14
 12.2 SAS Resources 12-11

Selected Additional Resources

Search Papers Presented at SAS Global Forum (previously known as SUGI)

• http://support.sas.com/events/sasglobalforum/previous/online.html

SAS Code Samples on support.sas.com

• http://support.sas.com/ctx/samples/index.jsp

SAS Code Samples from Specific Books

• http://support.sas.com/documentation/onlinedoc/code.samples.html

List of all SAS Products and Solutions

• http://www.sas.com/products/index.html

List of Papers
• http://support.sas.com/resources/papers/
12-12 Chapter 12 Learning More

12.3 Beyond This Course

Objectives
„ Identify the next set of courses that follow this course.

16

Next Steps
SAS® Programming 3:
Advanced Techniques
and Efficiencies
Applications
SAS Macro
Development
Language
Curriculum

Web Presenting
Enablement Your
Curriculum Data Statistical Information
Warehousing Analysis
Curriculum Curriculum

17
 12.3 Beyond This Course 12-13

Next Steps
To learn more about this: Enroll in the following:
SAS® Macro Language 1:
Essentials
Using the Macro
SAS® Macro Language 2:
Facility
Developing Macro
Applications

Creating tabular and SAS® Report Writing 1:

summary reports Using Procedures and ODS

Creating graphic
reports with SAS/GRAPH® 1: Essentials
SAS/GRAPH software

Processing data with

Structured Query SAS® SQL 1: Essentials
Language (SQL)
18

Next Steps
In addition, there are prerecorded, short, technical
discussions and demonstrations that are called e-lectures.

http://support.sas.com/training/

19
12-14 Chapter 12 Learning More
Appendix A Index
business scenario, 3-5, 3-50–3-51, 6-7–6-8,
% 6-68, 6-81, 7-4, 7-15, 7-25, 8-71–8-72, 9-
36–9-37, 9-43, 9-47
%INCLUDE statement, 10-87
BY statement
%SYSRC macro, 8-26
DESCENDING option, 9-36
_ GROUPFORMAT option, 9-48
indexes, 9-36
_FREQ_ variable NOTSORTED option, 9-36
SUMMARY procedure, 8-52 versus CLASS statement, 2-47–2-48, 9-51
_IORC_ automatic variable, 8-25–8-27, 8-35 BY-group processing, 9-33–9-52, 12-5
_TYPE_ variable CLASS statement, 9-33
SUMMARY procedure, 8-52 indexes, 9-33–9-36
NOTSORTED option, 9-33
A SORT procedure, 9-33
additional information user-sort assertion, 9-33
links to, 12-6–12-7, 12-11 BYSORTED system option, 9-46
alignment, 10-27
AND operator, 3-35 C
APPEND procedure, 3-43 CALCULATED keyword, 8-60
ARRAY statement, 4-7 CALL MISSING statement, 6-26
one-dimensional arrays, 5-6–5-7 CASE_FIRST= suboption
syntax, 4-6 values, 9-25
arrays, 6-8 CAT function, 10-8
advantages of using, 5-61 CATALOG procedure, 7-10
comparing with hash objects and formats, syntax, 7-11
7-33 catalogs, 7-8
disadvantages of using, 5-61 FMTSEARCH= system option, 7-15
multidimensional, 5-22–5-28, 5-41–5-60 CATQ function, 10-8
one-dimensional, 5-3–5-16 CATS function, 10-8
overview, 4-6 CATT function, 10-8
versus hash objects, 6-41 CATX function, 10-8
assignment statement, 10-18 CEIL function, 3-60
PUT function, 7-12 centiles, 3-38
attributes, 6-6 CENTILES option, 3-38
AUTOCALL library, 8-26 chained lookups
autoexec files, 11-4–11-6, 11-22–11-26 using hash objects, 6-67–6-83
disabling, 11-27 CLASS statement, 9-50–9-51, 10-68, 12-5
ECHOAUTO system option, 11-26 BY-group processing, 9-33
NOAUTOEXEC system option, 11-27 versus BY statement, 2-47–2-48, 9-51
CLOSE value
B SASFILE statement, 2-12
BEST. format, 10-8 CMPLIB= system option, 10-102
BUFNO= system option, 2-57 CNTLIN= option, 7-6, 7-16
FORMAT procedure, 7-6
CNTLOUT= option, 7-6, 7-16
A-2 Index

variables, 7-18 versus SURVEYSELECT procedure, 3-69

colon modifier, 3-30 DATA step arrays, 5-3–5-61
combining data conditionally, 8-73–8-91 DATA step merges, 8-4
combining data horizontally, 4-13, 8-3–8-91 advantages, 8-8
COMPARE procedure, 2-20–2-21 disadvantages, 8-8
compound optimization, Error! Not a valid versus PROC SQL joins, 8-12–8-16
bookmark in entry on page 3-35 DATA step views
compressing data sets, 2-30–2-42 compared to SAS data sets, 10-52–10-53
trade-offs, 2-41–2-42 comparing to SQL views, 10-71
concatenation functions, 10-8 creating, 10-54–10-58, 10-72
CONFIG= invocation system option DATASETS procedure, 3-17, 3-21–3-23, 3-
z/OS, 11-6 26, 3-38, 3-43, 9-9
configuration files, 11-4–11-6 syntax, 3-22
defining, 11-7–11-21 DECLARE statement, 6-16
environment variables, 11-12 hash objects, 6-6, 6-13
SAS librefs, 11-12 hiter objects, 6-49
UNIX, 11-13, 11-20 DEFAULT= option
Windows, 11-13, 11-21 LENGTH statement, 2-19
z/OS, 11-14–11-18, 11-21 DESCENDING option, 9-36
CONTENTS procedure, 3-26, 3-38, 9-9 DESCRIBE statement
reporting page size, 1-30 DATA step, 10-60
CPORT procedure, 3-44 SQL procedure, 10-61
CPUCOUNT= system option, 9-7 syntax, 10-60
CREATE INDEX statement, 3-25 detail data
customizing a SAS session, 11-4–11-6 combining with summary data, 8-50–8-63
discriminating variable, 3-45
D disk storage techniques, 4-13–4-26
DO loops, 3-51
data files
DO UNTIL statement, 10-21
compressing, 2-30–2-42
DOWNLOAD procedure, 3-44
data set page
duplicate key values, 8-36–8-37
definition, 1-29
DUPOUT= option
data sets
PROC SORT statement, 9-21–9-22
combining conditionally, 8-73–8-91
combining using multiple SET
E
statements, 8-24
compared to DATA step views, 10-52–10- ECHOAUTO system option, 11-26
53 efficiency, 1-12–1-15
comparing, 2-20–2-21 EMAIL option, 10-87
compressing, 2-30–2-42 ENABLEDIRECTIO option
reducing size, 12-4 LIBNAME statement, 2-57–2-58
sorting, 9-3–9-52 END= option, 10-21
summary, 8-51 ENDSUB statement
DATA statement FCMP procedure, 10-100
VIEW= option, 10-58 entries, 7-8
DATA step, 3-32, 3-50–3-52, 8-62–8-63 EXCLUDE statement, 7-12
combining summary and detail data, 8-53 FORMAT procedure, 7-17
hash objects, 6-4–6-5 EXecute Channel Program (EXCP), 1-28
macro variables, 10-73
match-merges, 8-7 F
sending e-mail, 10-87–10-88
FCMP procedure, 10-95
 Index A-3

advantages, 10-108 FUNCTION statement

calling functions created by, 10-100 FCMP procedure, 10-98
creating functions, 10-96
creating subroutines, 10-109–10-112 G
disadvantages, 10-108
GROUPFORMAT option, 9-48
ENDSUB statement, 10-100
advantages, 9-49
FUNCTION statement, 10-98
functions, 10-100–10-107 H
RETURN statement, 10-99
using functions created by, 10-101–10-102 hash objects, 6-4–6-5
FILE statement, 10-87 advantages, 6-41
advantages for writing code, 10-89 argument tags, 6-14–6-15
FILENAME statement, 10-12–10-14, 10-87 attributes, 6-13
syntax, 10-14 combining data conditionally, 8-90
FILEVAR= option comparing with formats and arrays, 7-33
INFILE statement, 10-15–10-16, 10-18 DECLARE statement, 6-6, 6-13
FIND method, 4-9 declaring, 6-13
FIRST() method, 6-49 lookup values, 6-8
FLOOR function, 3-52 methods, 6-13, 6-17–6-18
FMTLIB option overview, 4-9
PROC FORMAT statement, 7-12 SUM method, 8-63
FMTSEARCH= system option, 7-14–7-15 using for chained lookups, 6-67–6-83
FORMAT procedure, 7-3 versus arrays, 6-41
CNTLIN= option, 7-6, 7-16 hiter objects, 6-13, 6-50–6-62
CNTLOUT= option, 7-16 overview, 6-48
EXCLUDE statement, 7-17 selected methods, 6-49
maintaining permanent formats, 7-16 HOST option, 1-21
PICTURE statement, 7-27–7-33 host sort utilities, 9-12
SELECT statement, 7-17
FORMAT statement, 7-4, 7-12 I
FORMAT= option, 7-12 I/O, 3-7
formats, 7-12–7-13 direct file, 2-57–2-58
advantages of using, 7-19 factors affecting, 3-39
comparing with hash objects and arrays, reducing operations, 2-5–2-12, 12-3
7-33 IDXWHERE= option, 3-40
creating using a control data set, 7-5–7-6 IF statement
disadvantages of using, 7-19 PUT function, 7-12
documenting, 7-9–7-11 IF-THEN statement, 6-41
FMTSEARCH= system option, 7-15 IN operator, 3-35
maintaining, 7-16–7-18, 7-16–7-18 INDEX CREATE statement, 3-22
nesting, 7-8 INDEX= data set option, 3-44
overview, 4-11–4-12, 7-4 indexes, 3-19
permanent, 7-16–7-18 indexes, 3-4–3-26, 3-29–3-46, 8-25
storing, 7-8 BY statement, 9-36
user-defined, 4-11 BY-group processing, 9-33–9-36
FULLSTIMER option, 1-20–1-21, 6-15 centiles, 3-38
FULLSTIMER statistics comparing creation techniques, 3-26
Window, 1-25 data order, 3-39
z/OS, 1-28 documenting, 3-26
function definitions INDEX= data set option, 3-19
FCMP procedure, 10-98 maintaining, 3-43–3-46
A-4 Index

not used, 3-32 M

purpose, 3-5
macro facility, 10-5–10-6
reading data sets with, 3-7
MACRO system option, 8-26
subset size, 3-37–3-38
macro variables, 10-73
UNIQUE option, 9-39–9-40
match-merging
INFILE statement, 1-31, 10-12
using a PROC SQL join, 8-10
FILEVAR= option, 10-15–10-16, 10-18
using the DATA step, 8-7
inner joins
MEANS procedure, 10-68
SQL procedure, 4-17, 8-9
CLASS statement, 9-50–9-51
INPUT statement, 1-31, 10-18
multi-threaded processing, 9-5
INSERT INTO statement, 3-43
SUMSIZE= option, 9-52
instantiated, 6-13
memory
INTNX function, 10-25–10-27, 10-25–10-29
reducing usage, 12-5
syntax, 10-26, 10-104
MEMRPT option, 1-20–1-21
using, 10-104
MEMSIZE= system option, 6-15
IORCMSG function, 8-26
MERGE statement, 1-32, 3-43
combining data conditionally, 8-76
J
merging
joining tables, 4-17 overview, 4-13–4-15
using the DATA step, 8-4, 8-8
K versus SQL inner joins, 8-12–8-16
KEY= option, 8-24–8-26 METHOD= option, 3-68
MODIFY statement, 8-25 methods, 6-6
SET statement, 8-25 MODIFY statement
KEY= option, 8-25
L MONTH function, 10-23
MSGLEVEL= system option, 3-20–3-21, 3-
LAST() method, 6-49 24
LENGTH statement, 6-41 multidimensional arrays, 5-22–5-28
DEFAULT= option, 2-19 loading from SAS data sets, 5-41–5-60
LIBNAME statement multiple SET statements, 4-18–4-25
ENABLEDIRECTIO option, 2-57–2-58
USEDIRECTIO= option, 2-57–2-58 N
LIBRARY= option
PROC FORMAT statement, 7-9 network bandwidth, 1-11
LINGUISTIC option, 9-24 NEXT() method, 6-49
CASE_FIRST= suboption, 9-25 NOAUTOEXEC system option, 11-27
STRENGTH= suboption, 9-26 NOBS= option, 3-50–3-52
LIST option, 11-38 NODUPKEY option
LISTUSER option, 11-37 PROC SORT statement, 9-21–9-22
LOAD value NOMISS option, 3-17–3-19, 3-22, 3-25
SASFILE statement, 2-12 NOSGIO option, 2-56
LOCALE system option, 9-24 NOTHREADS option
lookup techniques PROC SORT statement, 9-6
in-memory, 4-6–4-11 NOTHREADS system option, 9-6
overview, 4-4 NOTSORTED option, 9-36, 9-44–9-46, 9-48
lookup values BY-group processing, 9-33
hash objects, 6-8 numeric variables
tables, 4-3 default length, 2-17
NUMERIC_COLLATION= option
PROC SORT statement, 9-28–9-29
 Index A-5

O SORTSIZE= option, 9-10

THREADS option, 9-6
object dot syntax, 6-17
PROC SQL joins
one-dimensional arrays, 5-8–5-16
advantages, 8-11
ARRAY statement, 5-6–5-7
disadvantages, 8-11
OPEN value
using to match-merge, 8-10
SASFILE statement, 2-12
versus DATA step merges, 8-12–8-16
operators
program data vector (PDV), 1-31
AND, 3-35
program resources, 1-11
IN, 3-35
PRTDEF procedure, 11-31
SOUNDS-LIKE, 3-34
PUT function, 7-4, 7-12
OPTIONS procedure, 1-21, 11-3–11-4
PUT statement, 7-4, 7-12, 10-87
OPTIONS statement, 1-21
advantages for writing code, 10-89
OUT= option, 3-44
PUTLOG statement, 8-27
OUTCAT= option
PROC FCMP statement, 10-97
R
OUTLIB= option
PROC FCMP statement, 10-97 random sample, 3-58, 3-64
Output Delivery System. See ODS RANUNI function, 3-58
OUTPUT statement, 3-51, 10-18 raw data
SUMMARY procedure, 8-53 reading, 10-17, 10-21
OVERWRITE option REGISTRY procedure, 11-32, 11-36–11-38
PROC SORT statement, 9-9 LIST option, 11-38
LISTUSER option, 11-37
P STARTAT= option, 11-38
REPORT procedure
PDV variables, 6-41
multi-threaded processing, 9-5
permanent formats
SUMSIZE= option, 9-52
maintaining, 7-16–7-18
resources
picture formats, 7-24–7-30
usage, 1-20–1-21
PICTURE statement, 7-27–7-33
RETURN statement
DATATYPE= option, 7-33
FCMP procedure, 10-99
options, 7-31
POINT= option, 3-52
S
PRESORTED option
PROC SORT statement, 9-17, 10-65 SAS
PREV() method, 6-49 customizing a session, 11-4–11-6
PRINT procedure SAS catalogs, 7-8
UNIFORM option, 10-68 SAS configuration files
PROC FCMP statement defining, 11-7–11-21
syntax, 10-97 SAS Explorer, 3-26, 3-43
PROC FORMAT statement SAS Management Console, 3-26
FMTLIB option, 7-12 SAS Registry, 11-4–11-6, 11-28–11-38
LIBRARY= option, 7-9 LIBNAME keys, 11-33
PROC SORT statement New Library window, 11-30
DUPOUT= option, 9-21–9-22 Print Setup window, 11-30
NODUPKEY option, 9-21–9-22 printer keys, 11-34
NOTHREADS option, 9-6 Registry Editor, 11-32
NUMERIC_COLLATION= option, 9-28– REGISTRY procedure, 11-32, 11-36–11-38
9-29 SAS Explorer, 11-31
OVERWRITE option, 9-9 SAS Explorer keys, 11-35
PRESORTED option, 9-17, 10-65 storage location, 11-29
A-6 Index

techniques for modifying the syntax, 3-25

configuration, 11-28–11-38 SQL views, 10-61–10-62
SAS sort utility, 9-4 comparing to DATA step views, 10-71
SAS/CONNECT, 3-44 STARTAT= option, 11-38
SAS/STAT, 3-65, 10-68 STATS option, 1-20–1-21
SASFILE statement, 2-11–2-12 STIMER option, 1-20–1-21
seed, 3-58 STOP statement, 3-51, 3-53, 6-64, 10-18
SEED= option, 3-69 stored array values, 5-41
SELECT statement, 7-12, 8-60 STRENGTH= suboption
FORMAT procedure, 7-17 values, 9-26
SET statement, 3-43, 3-52, 8-62 subsetting IF statement, 3-32
KEY= option, 8-25, 8-24–8-26 SUBSTR function, 3-34
reading a SAS data set, 1-32 SUM function, 8-60
using multiple in the DATA step, 4-18–4- SUM method
25, 8-76 hash object, 8-63
SGIO option, 2-56–2-57 summary data
sort indicator, 9-14 combining with detail data, 8-50–8-63
SORT procedure summary data sets
BY-group processing, 9-33 creating, 8-51
collating sequence, 9-19, 9-23 SUMMARY procedure, 8-52, 10-68
multi-threaded processing, 9-4–9-5 _FREQ_ variable, 8-52
sort order, 9-19 _TYPE_ variable, 8-52
sort space requirements, 9-29 CLASS statement, 9-50–9-51
TAGSORT option, 9-6 multi-threaded processing, 9-5
sort space OUTPUT statement, 8-53
requirements, 9-29 SUMSIZE= option, 9-52
sort validation, 9-14 SUMSIZE= option, 9-52
SORTCUTP= system option, 9-12 SURVEYSELECT procedure, 3-65–3-69
SORTEDBY= option, 9-15–9-16 METHOD= option, 3-68
sorting data sets, 9-3–9-52 SEED= option, 3-69
reasons for sorting, 9-3 versus DATA step, 3-69
SORTNAME= system option, 9-12 SYMGET function, 10-73
SORTPGM= system option, 9-12
SORTSIZE= option, 9-4 T
PROC SORT statement, 9-10
TABULATE procedure, 8-61, 10-68
SORTVALIDATE system option, 9-17
CLASS statement, 9-50–9-51
SOUNDS-LIKE operator, 3-34
multi-threaded processing, 9-5
SQL inner joins
SUMSIZE= option, 9-52
versus mergin, 8-12–8-16
tags
SQL procedure, 3-17, 3-24–3-25, 3-43
observation numbers, 9-6
colon modifier, 3-30
TAGSORT option, 9-6, 9-9
combining data conditionally, 8-89
THREADS option
combining detail data and summary data,
PROC SORT statement, 9-6
8-59–8-61
THREADS system option, 9-6
creating views, 10-61–10-62
TRIM function, 3-34
DESCRIBE statement, 10-61
inner joins, 4-17, 8-9 U
joins, 8-10
match-merging, 8-10 UNIFORM function, 3-58
multi-threaded processing, 9-5 UNIQUE option, 3-17–3-19, 3-22, 3-25, 8-37
remerging data, 8-60 KEY= option, 9-39–9-40
 Index A-7

UNIVARIATE procedure, 10-68 views

CLASS statement, 9-50–9-51 creating using the SQL procedure, 10-61–
UPDATE statement, 3-43 10-62
UPLOAD procedure, 3-44 DATA step, 10-54–10-58, 10-72
USEDIRECTIO= option DATA step versus SQL, 10-71
LIBNAME statement, 2-57–2-58 guidelines, 10-65–10-69
user-defined formats
syntax, 4-11 W
user-sort assertion
WHERE expressions, 3-32–3-37
BY-group processing, 9-33
WHERE statement, 1-28
PUT function, 7-12
V
WIDTH= option
variables PROC PRINT statement, 10-69
discriminating, 3-45 -WORK system option
VIEW= option z/OS, 11-19
DATA statement, 10-58
 Recommended SAS® Titles
SAS® Programming 3: Advanced Techniques and Efficiencies
Price
(U.S.
ISBN Title dollars)
SAS® Press
978-1-55544-806-6 An Array of Challenges—Test Your SAS® Skills $23.95
978-1-58025-578-3 Annotate: Simply the Basics $24.95
978-1-59994-659-7 Cody's Data Cleaning Techniques Using SAS®, Second Edition $39.95
978-1-59047-920-9 Combining and Modifying SAS® Data Sets: Examples, Second Edition $44.95
978-1-59047-849-3 The Complete Guide to SAS® Indexes $54.95
978-1-58025-927-9 Debugging SAS® Programs: A Handbook of Tools and Techniques $47.95
In the Know…SAS® Tips and Techniques From Around the Globe,
978-1-59047-702-1 $55.95
Second Edition
978-1-59994-649-8 Just Enough SAS®: A Quick-Start Guide to SAS® for Engineers $49.95
978-1-59994-165-3 Learning SAS® by Example: A Programmer's Guide $69.95
978-1-59994-725-9 The Little SAS® Book: A Primer, Fourth Edition $49.95
978-1-58025-924-8 Longitudinal Data and SAS®: A Programmer's Guide $29.95
978-1-891957-12-3 Professional SAS® Programmer's Pocket Reference, Fifth Edition $17.95
978-1-891957-11-6 Professional SAS® Programming Shortcuts, Second Edition $39.95
978-0-470-53968-2 SAS® For Dummies®, Second Edition $29.99
978-1-60764-340-1 SAS® Functions by Example, Second Edition $54.95
978-1-59047-793-9 SAS® Programming in the Pharmaceutical Industry $50.95
978-1-59047-574-4 Saving Time and Money Using SAS® $34.95
978-1-59047-149-4 Step-by-Step Basic Statistics Using SAS®: Exercises $54.95
978-1-59047-148-7 Step-by-Step Basic Statistics Using SAS®: Student Guide $74.95
978-1-59047-150-0 Step-by-Step Basic Statistics Using SAS®: Student Guide and Exercises $99.95
978-1-59047-573-7 The Power of PROC FORMAT $29.95
978-1-58025-660-5 Visualizing Categorical Data $69.95
SAS® Certification Prep Guide: Advanced Programming for SAS® 9,
978-1-60764-044-8 $129.00
Second Edition
978-1-60764-045-5 SAS® Certification Prep Guide: Base Programming for SAS® 9, Second Edition #129.00
978-1-60764-353-1 SAS® OnlineDoc 9.2: PDF Files, Second Edition (CD-ROM) #29.95

Notes
x Prices are subject to change without notice.
x SAS® 9 documentation is also available online at: support.sas.com/documentation
x To order, please visit: support.sas.com/bookstore