Software Project Management (Lecture 9)

Dr. R. Mall

Organization of this Lecture:

Introduction to Project Planning Software Cost Estimation
Cost Estimation Models Software Size Metrics Empirical Estimation Heuristic Estimation COCOMO

Staffing Level Estimation Effect of Schedule Compression on Cost Summary

2

Introduction
Many software projects fail:
due to faulty project management practices:
It is important to learn different aspects of software project management.

Introduction
Goal of software project management:
enable a group of engineers to work efficiently towards successful completion of a software project.

Responsibility of project managers

Project proposal writing, Project cost estimation, Scheduling, Project staffing, Project monitoring and control, Software configuration management, Risk management, Managerial report writing and presentations, etc.

Introduction
A project managers activities are varied.
can be broadly classified into:
project planning, project monitoring and control activities.
6

Project Planning
Once a project is found to be feasible,
project managers undertake project planning.

Project Planning Activities

Estimation:
Effort, cost, resource, and project duration

Project scheduling: Staff organization: Risk handling:

staffing plans

Miscellaneous plans:

identification, analysis, and abatement procedures quality assurance plan, configuration management plan, etc.

Project planning
Requires utmost care and attention --commitments to unrealistic time and resource estimates result in:
irritating delays. customer dissatisfaction adverse affect on team morale
poor quality work

project failure.
9

Sliding Window Planning

Involves project planning over several stages:

protects managers from making big commitments too early. More information becomes available as project progresses. Facilitates accurate planning
10

SPMP Document
After planning is complete:
Document the plans: in a Software Project Management Plan(SPMP) document.
11

Organization of SPMP Document

Introduction
Constraints)

(Objectives,Major Functions,Performance Issues,Management and Technical

Project Estimates (Historical Data,Estimation Techniques,Effort, Cost, and Project Duration Estimates) Project Resources Plan (People,Hardware and Software,Special Resources) Schedules (Work Breakdown Structure,Task Network, Gantt Chart Representation,PERT Chart
Representation)

Risk Management Plan

Procedures)

(Risk Analysis,Risk Identification,Risk Estimation, Abatement

Project Tracking and Control Plan Miscellaneous Plans(Process Tailoring,Quality Assurance)

12

Software Cost Estimation

Determine size of the product.
From the size estimate,
determine the effort needed.

From the effort estimate,

determine project duration, and cost.

13

Software Cost Estimation

Effort Estimation Size Estimation Cost Estimation

Staffing Estimation Duration Estimation

Scheduling

14

Software Cost Estimation

Three main approaches to estimation:

Empirical Heuristic Analytical
15

Software Cost Estimation Techniques

Empirical techniques:
Heuristic techniques:

an educated guess based on past experience.

Analytical techniques:

assume that the characteristics to be estimated can be expressed in terms of some mathematical expression.
derive the required results starting from certain simple assumptions.
16

Software Size Metrics

LOC (Lines of Code):
Simplest and most widely used metric. Comments and blank lines should not be counted.

17

Disadvantages of Using LOC

Size can vary with coding style. Focuses on coding activity alone. Correlates poorly with quality and efficiency of code. Penalizes higher level programming languages, code reuse, etc.
18

Disadvantages of Using LOC

(cont...)

Measures lexical/textual complexity only.

does not address the issues of structural or logical complexity.

Difficult to estimate LOC from problem description.

So not useful for project planning
19

Function Point Metric

Overcomes some of the shortcomings of the LOC metric
Proposed by Albrecht in early 80's:
FP=4 #inputs + 5 #Outputs + 4 #inquiries + 10 #files + 10 #interfaces Input:
A set of related inputs is counted as one input.

20

Function Point Metric

Output:
A set of related outputs is counted as one output.

Inquiries:
Each user query type is counted.

Files:
Files are logically related data and thus can be data structures or physical files.

Interface:
Data transfer to other systems.
21

Function Point Metric

(CONT.)

Suffers from a major drawback:

the size of a function is considered to be independent of its complexity.

Extend function point metric:

Feature Point metric: considers an extra parameter:
Algorithm Complexity.
22

Function Point Metric

Proponents claim:

(CONT.)

FP is language independent. Size can be easily derived from problem description

Opponents claim:
it is subjective --- Different people can come up with different estimates for the same problem.
23

Empirical Size Estimation Techniques

Expert Judgement:

Delphi Estimation:

An euphemism for guess made by an expert. Suffers from individual bias.

overcomes some of the problems of expert judgement.

24

Expert judgement
Experts divide a software product into component units:
e.g. GUI, database module, data communication module, billing module, etc.

Add up the guesses for each of the components.

25

Delphi Estimation:
Team of Experts and a coordinator. Experts carry out estimation independently:
mention the rationale behind their estimation. coordinator notes down any extraordinary rationale:
circulates among experts.
26

Delphi Estimation:
Experts re-estimate. Experts never meet each other
to discuss their viewpoints.

27

Heuristic Estimation Techniques

Single Variable Model:

Parameter to be Estimated=C1(Estimated Characteristic)d1

Multivariable Model:

Assumes that the parameter to be estimated depends on more than one characteristic.

Parameter to be Estimated=C1(Estimated Characteristic)d1+ C2(Estimated Characteristic)d2+

Usually more accurate than single variable models.

28

COCOMO Model
COCOMO (COnstructive COst MOdel) proposed by Boehm. Divides software product developments into 3 categories:
Organic Semidetached Embedded
29

COCOMO Product classes

Roughly correspond to:
application, utility and system programs respectively.
Data processing and scientific programs are considered to be application programs. Compilers, linkers, editors, etc., are utility programs. Operating systems and real-time system programs, etc. are system programs.
30

Elaboration of Product classes

Organic:
Relatively small groups
working to develop well-understood applications.

Semidetached:

Project team consists of a mixture of experienced and inexperienced staff. The software is strongly coupled to complex hardware, or real-time systems.

Embedded:

31

COCOMO Model

(CONT.)

For each of the three product categories:

From size estimation (in KLOC), Boehm provides

equations to predict:

project duration in months effort in programmer-months

Boehm obtained these equations:

examined historical data collected from a large number of actual projects.
32

COCOMO Model

(CONT.)

Software cost estimation is done through three stages:

Basic COCOMO, Intermediate COCOMO, Complete COCOMO.

33

Basic COCOMO Model

Gives only an approximate estimation: Effort = a1 (KLOC)a2 Tdev = b1 (Effort)b2

(CONT.)

KLOC is the estimated kilo lines of source code, a1,a2,b1,b2 are constants for different categories of software products, Tdev is the estimated time to develop the software in months, Effort estimation is obtained in terms of 34

Development Effort Estimation

Organic :
Effort = 2.4 (KLOC)1.05 PM

Semi-detached:
Effort = 3.0(KLOC)1.12 PM

Embedded:
Effort = 3.6 (KLOC)1.20PM
35

Development Time Estimation

Organic:
Tdev = 2.5 (Effort)0.38 Months

Semi-detached:
Tdev = 2.5 (Effort)0.35 Months

Embedded:
Tdev = 2.5 (Effort)0.32 Months
36

Basic COCOMO Model

(CONT.)

Effort Effort is somewhat super-linear in problem size.

Size

37

Basic COCOMO Model

Development time
sublinear function of Dev. Time product size.

(CONT.)

When product size increases two times,

development time does not double.

18 Months 14 Months

Time taken:
almost same for all the three product categories.

Size

30K

60K

38

Basic COCOMO Model

(CONT.)

Development time does not increase linearly with product size:

For larger products more parallel activities can be identified:
can be carried out simultaneously by a number of engineers.

39

Basic COCOMO Model

(CONT.)

Development time is roughly the same for all the three categories of products:
For example, a 60 KLOC program can be developed in approximately 18 months
regardless of whether it is of organic, semidetached, or embedded type.

There is more scope for parallel activities for system and application programs,
than utility programs.
40

Example
The size of an organic software product has been estimated to be 32,000 lines of source code.
Effort = 2.4*(32)1.05 = 91 PM Nominal development time = 2.5*(91)0.38 = 14 months

41

Intermediate COCOMO
Basic COCOMO model assumes
effort and development time depend on product size alone.

However, several parameters affect effort and development time:

Reliability requirements Availability of CASE tools and modern facilities to the developers Size of data to be handled
42

Intermediate COCOMO
For accurate estimation,
the effect of all relevant parameters must be considered: Intermediate COCOMO model recognizes this fact:
refines the initial estimate obtained by the basic COCOMO by using a set of 15 cost drivers (multipliers).
43

Intermediate COCOMO
(CONT.)

If modern programming practices are used,

initial estimates are scaled downwards.

If there are stringent reliability requirements on the product :

initial estimate is scaled upwards.
44

Intermediate COCOMO
(CONT.)

Rate different parameters on a scale of one to three:

Depending on these ratings,
multiply cost driver values with the estimate obtained using the basic COCOMO.
45

Intermediate COCOMO
(CONT.)

Cost driver classes:

Product: Inherent complexity of the product, reliability requirements of the product, etc. Computer: Execution time, storage requirements, etc. Personnel: Experience of personnel, etc. Development Environment: Sophistication of the tools used for software development.
46

Shortcoming of basic and intermediate COCOMO models

Both models:
consider a software product as a single homogeneous entity: However, most large systems are made up of several smaller sub-systems.
Some sub-systems may be considered as organic type, some may be considered embedded, etc. for some the reliability requirements may be high, and so on.
47

Complete COCOMO
Cost of each sub-system is estimated separately. Costs of the sub-systems are added to obtain total cost. Reduces the margin of error in the final estimate.
48

Complete COCOMO Example

A Management Information System (MIS) for an organization having offices at several places across the country:
Database part (semi-detached) Graphical User Interface (GUI) part (organic) Communication part (embedded)

Costs of the components are estimated separately:

summed up to give the overall cost of the system.
49

Halstead's Software Science

An analytical technique to estimate:
size, development effort, development time.

50

Halstead's Software Science

Halstead used a few primitive program parameters
number of operators and operands

Derived expressions for:

over all program length, potential minimum volume actual volume, language level, effort, and development time.
51

Staffing Level Estimation

Number of personnel required during any development project:
not constant.

Norden in 1958 analyzed many R&D projects, and observed:

Rayleigh curve represents the number of full-time personnel required at any time.
52

Rayleigh Curve
Rayleigh curve is specified by two parameters: Effort
td the time at which the curve reaches its maximum K the total area under the curve. Rayleigh Curve

td

Time

L=f(K, td)
53

Putnams Work:
In 1976, Putnam studied the problem of staffing of software projects:
observed that the level of effort required in software development efforts has a similar envelope. found that the Rayleigh-Norden curve
relates the number of delivered lines of code to effort and development time.

54

Putnams Work

(CONT.)

Putnam analyzed a large number of army projects, and derived the expression: L=CkK1/3td4/3

K is the effort expended and L is the size in KLOC. td is the time to develop the software. Ck is the state of technology constant
reflects factors that affect programmer productivity.
55

Putnams Work

(CONT.)

Ck=2 for poor development environment

no methodology, poor documentation, and review, etc.

Ck=8 for good software development environment

software engineering principles used

Ck=11 for an excellent environment

56

Rayleigh Curve
Very small number of engineers are needed at the beginning of a project
carry out planning and specification.

As the project progresses:

more detailed work is required, number of engineers slowly increases and reaches a peak.
57

Rayleigh Curve
Putnam observed that:
the time at which the Rayleigh curve reaches its maximum value
corresponds to system testing and product release.

After system testing,

the number of project staff falls till product installation and delivery.
58

Rayleigh Curve
From the Rayleigh curve observe that:
approximately 40% of the area under the Rayleigh curve is to the left of td and 60% to the right.
59

Effect of Schedule Change on Cost

Using the Putnam's expression for L, K=L3/Ck3td4 Or, K=C1/td4 For the same product size, C1=L3/Ck3 is a constant. Or, K1/K2 = td24/td14
60

Effect of Schedule Change on

Cost

(CONT.)

Observe:
a relatively small compression in delivery schedule
can result in substantial penalty on human effort.

Also, observe:
benefits can be gained by using fewer people over a somewhat longer time span.

61

Example
If the estimated development time is 1 year, then in order to develop the product in 6 months,
the total effort and hence the cost increases 16 times. In other words,
the relationship between effort and the chronological delivery time is highly nonlinear.

62

Effect of Schedule Change on Cost

(CONT.)

Putnam model indicates extreme penalty for schedule compression

and extreme reward for expanding the schedule.

Putnam estimation model works reasonably well for very large systems,
but seriously overestimates the effort for medium and small systems.
63

Effect of Schedule Change on Cost

(CONT.)

Boehm observed:
There is a limit beyond which the schedule of a software project cannot be reduced by buying any more personnel or equipment. This limit occurs roughly at 75% of the nominal time estimate.
64

Effect of Schedule Change on Cost

(CONT.)

If a project manager accepts a customer demand to compress the development time by more than 25%
very unlikely to succeed.
every project has only a limited amount of parallel activities sequential activities cannot be speeded up by hiring any number of additional engineers. many engineers have to sit idle.
65

Jensen Model
Jensen model is very similar to Putnam model.
attempts to soften the effect of schedule compression on effort makes it applicable to smaller and medium sized projects.
66

Jensen Model
Jensen proposed the equation:
L=CtetdK1/2 Where,
Cte is the effective technology constant, td is the time to develop the software, and K is the effort needed to develop the software.
67

Organization Structure
Functional Organization:
Engineers are organized into functional groups, e.g.
specification, design, coding, testing, maintenance, etc.

Engineers from functional groups get assigned to different projects

68

Advantages of Functional Organization Specialization Ease of staffing Good documentation is produced

different phases are carried out by different teams of engineers.

Helps identify errors earlier.

69

Project Organization
Engineers get assigned to a project for the entire duration of the project
Advantages:

Same set of engineers carry out all the phases Engineers save time on learning details of every project. Leads to job rotation
70

Team Structure
Problems of different complexities and sizes require different team structures:
Chief-programmer team Democratic team Mixed organization
71

Democratic Teams
Suitable for:
small projects requiring less than five or six engineers research-oriented projects

A manager provides administrative leadership:

at different times different members of the group provide technical leadership.
72

Democratic Teams
Democratic organization provides
higher morale and job satisfaction to the engineers therefore leads to less employee turnover.

Suitable for less understood problems,

a group of engineers can invent better solutions than a single individual.

73

Democratic Teams Disadvantage:

team members may waste a lot time arguing about trivial points:
absence of any authority in the team.
74

Chief Programmer Team

A senior engineer provides technical leadership:

partitions the task among the team members. verifies and integrates the products developed by the members.
75

Chief Programmer Team

Works well when
the task is well understood

importance of early completion outweighs other factors

also within the intellectual grasp of a single individual,

team morale, personal development, etc.

76

Chief Programmer Team

Chief programmer team is subject to single point failure:
too much responsibility and authority is assigned to the chief programmer.

77

Mixed Control Team Organization

Draws upon ideas from both:
democratic organization and chief-programmer team organization.

Communication is limited
to a small group that is most likely to benefit from it.

Suitable for large organizations.

78

Team Organization

Democratic Team Chief Programmer team

79

Mixed team organization

80

Summary
We discussed the broad responsibilities of the project manager:
Project planning Project Monitoring and Control

81

Summary
To estimate software cost:
Determine size of the product. Using size estimate,
determine effort needed.

From the effort estimate,

determine project duration, and cost.

82

Summary

(CONT.)

Cost estimation techniques:

Empirical Techniques Heuristic Techniques Analytical Techniques

Empirical techniques:
based on systematic guesses by experts.
Expert Judgement Delphi Estimation
83

Summary

(CONT.)

Heuristic techniques:
assume that characteristics of a software product can be modeled by a mathematical expression. COCOMO

Analytical techniques:
derive the estimates starting with some basic assumptions: Halstead's Software Science
84

Summary

(CONT.)

The staffing level during the life cycle of a software product development:
follows Rayleigh curve maximum number of engineers required during testing.

85

Summary

(CONT.)

Relationship between schedule change and effort:

highly nonlinear.

Software organizations are usually organized in:

functional format project format

86

Summary

(CONT.)

Project teams can be organized in following ways:

Chief programmer: suitable for routine work. Democratic: Small teams doing R&D type work Mixed: Large projects

87