Software Project

Management (Lecture 9)

Dr. R. Mall

1
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.

3
Introduction
 Goal of software project
management:
 enable a group of engineers to work
efficiently towards successful
completion of a software project.

4
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.

5
Introduction
 A project manager’s 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.

7
Project Planning Activities

 Estimation:
 Effort, cost, resource, and project duration
 Project scheduling:
 Staff organization:
 staffing plans
 Risk handling:
 identification, analysis, and abatement
procedures
 Miscellaneous plans:
 quality assurance plan, configuration
management plan, etc.
8
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 (Objectives,Major Functions,Performance Issues,Management and Technical

Constraints)

 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 (Risk Analysis,Risk Identification,Risk Estimation, Abatement

Procedures)

 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 Cost
Estimation Estimation

Size Staffing
Estimation Estimation

Duration
Estimation Scheduling

14
Software Cost Estimation

 Three main approaches to

estimation:
 Empirical
 Heuristic
 Analytical

15
Software Cost Estimation
Techniques

 Empirical techniques:
 an educated guess based on past experience.
 Heuristic techniques:
 assume that the characteristics to be
estimated can be expressed in terms of
some mathematical expression.
 Analytical techniques:
 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 (CONT.)

 Proponents claim:
 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:
 An euphemism for guess made by
an expert.
 Suffers from individual bias.
 Delphi Estimation:
 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.
 Embedded:
 The software is strongly coupled to complex
hardware, or real-time systems.

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 (CONT.)

 Gives only an approximate estimation:

 Effort = a1 (KLOC)a2
 Tdev = b1 (Effort)b2
 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
person months (PMs).
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.)

d
che
 Effort is Effort
em
id
et
a

S
somewhat be
d ded

super-linear in Em
Or
ga
n ic

problem size.
Size

37
Basic COCOMO Model (CONT.)

 Development time
 sublinear function of
product size. Dev. Time
 When product size ded ched
bed ide ta
increases two times, 18 Months Em Sem
 development time does
14 Months
not double. gan
ic
Or
 Time taken:
 almost same for all the 30K 60K
three product categories. Size

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, semi-
detached, 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 Rayleigh Curve

specified by two
parameters: Effort
 td the time at
which the curve
reaches its
maximum td
 K the total area Time
under the curve.
 L=f(K, td)
53
Putnam’s 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
Putnam’s 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
Putnam’s 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
 Same set of engineers carry out all the
phases
 Advantages:
 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
 also within the intellectual grasp of a single
individual,
 importance of early completion
outweighs other factors
 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