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