Estimation

Software cost estimation is the process of predicting the effort required to develop a software
system. Software measurement, just like any other measurement in the physical world, can be
categorized into Direct measures and Indirect measures.

Direct measures of the software process include measurement of cost and effort applied. Direct
measures of a product also include Line Of Code (LOC) produced, execution speed, memory
size, and defects reported over some of times.

Indirect measures of the software process include measurement of resulting features of the
software. Indirect measures of the product also include its functionality, quality, complexity,
usability etc.

On the other word, estimation is not an exact science. Too many variables – human, technical,
and environment – can affect the ultimate cost of software and effort applied to develop it.

In order to make a reliable cost and effort estimation, there are a lot of techniques that provide
estimates with acceptable amount of risk.

These methods can be categorized into-

a) Decomposition techniques :
In software cost estimation, decomposition techniques are widely used to break down a project into smaller, more
manageable parts to estimate size, effort, and cost more accurately.

I) Lines Of Code (LOC)

II) Functional Point (FP) Estimation

b) Empirical Estimation models:

Empirical Estimation Models are techniques used in software engineering to predict the effort, time, and cost required for a
project based on data from past projects. These models use formulas derived from historical data to make these predictions

I) COnstructive COst MOdel (COCOMO)

1. Basic COCOMO
2. Intermediate COCOMO
3. Advanced COCOMO

Line Of Code (LOC)

LOC is a direct approach method and requires a higher level of detail by means of decomposition
and partitioning. From statement of software scope, software is decomposed into problem
functions that can each be estimated individually. LOC is then calculated for each function.
An expected value is then computed using the following formula-

EV = ( Sopt + 4Sm + Spess) / 6

where
EV stand for the Estimation Variable
Sopt stand for Optimistic Estimation case when everything goes right
Sm stand for the most likely Estimation case given normal problems and opportunities
Spess stand for the pessimistic Estimation case where everything goes wrong

Once the expected value for estimation variable has been determined, historical LOC or FP data
are applied and person months, costs etc. are calculated using the following formula

Productivity = KLOC / Person-month

Quality = Defects / KLOC
Cost = $/LOC
Documentation = pages of documentation / KLOC

Where,
KLOC stand for number of code (in thousands)
Person-month stand for is the time (in months) taken by developers to finish the product
Defects stand for Total Number of errors discovered.

Example:
Problem Statement: Take our Library management system case. Software developed for library
will accept data from operator for issuing and returning books. Issuing and returning will require
some validity checks. For issue it is required to check if the member has already issued the
maximum books allowed. In case for return, if the member is returning the book after the due
date then fine has to be calculated. All the interactions will be through user interface. Other
operations include maintaining database and generating reports at regular intervals.

Major software functions identified

1. User interface
2. Database management
3. Report generation

For user interface

Sopt : 1800
Sm :2000
Spess : 4000

EV for user interface

EV = (1800 + 4*2000 + 4000) / 6
EV = 2300

For dababase managemant

Sopt : 4600
Sm : 6900
Spess : 8600

EV for database management

EV = (4600 + 4*6900 + 8600) / 6
EV = 6800

For report generation

Sopt : 1200
Sm : 1600
Spess : 3200

EV for database management

EV = (1200 + 4*1600 + 3200) / 6
EV = 1800

Function Estimated LOC (EV)

User interface 2300
Database management 6800
Report generation 1800
Estimated LOC 10900

Suppose productivity = 3000 LOC / person-month

Development Cost = $8000 person-month

Person-month = LOC / Productivity

= 10900/3000
= 3.6
Cost of software = Development Cost * person-month
= 8000 * 3.6
= 29120
Cost = Cost of software ($) / LOC
= 29120 / 10900
= 2.67 $ / LOC

FP Estimation
Function oriented metrics focus on program “functionality” or “utility”. Albrecht first proposed
function point method, which is a function oriented productivity measurement approach.
Five information domain characteristics are determined and counts for each are provided and
recorded in a table.

 Number of user inputs

 Number of user outputs
 Number of user inquires
 Number of files
 Number of external interfaces

Once the data have been collected, a complexity value is associated with each count. Each entry
can be simple, average or complex. Depending upon these complexity values is calculated.
To compute function points, we use

FP = count-total X [0.65 + 0.01 * SUM (Fi)]

Fi (I= 1 to 14) are complexity adjustment values based on response to questions (1-14) given
below. The constant values in the equation and the weighing factors that are applied to
information domain are determined empirically.
Fi
1. Does the system require reliable backup and recovery?
2. Are data communications required?
3. Are there distributed processing functions?
4. Is performance critical?
5. Will the system run in an existing, heavily utilized operational environment?
6. Does the system require on-line entry?
7. Does the on-line data entry require the input transaction to be built over multiple screens or
operations?
8. Are the inputs, files, or inquires complex
9. Is the internal processing complex?
10. Is the code designed to be reusable?
11. Are masters files updated online?
12. Are conversation and installations included in the design?
13. Is the system designed for multiple installations in different organizations?
14. Is the application designed to facilitate change and ease of use by the user?

Rate each factor on a scale of 0 to 5

0 – No influence
1 – Incidental
2 – Moderate
3 – Average
4 – Significant
5 – Essential
Count-total is sum of all FP entries.
Once functional point has been calculated, productivity, quality, cost, and documentation can be
evaluated.

Productivity = FP / person-month
Quality = defects / FP
Cost = $ / FP
Documentation = pages of documentation / FP

Example:

Considering the following information for a software development project to answer the question
given below-

Same as LOC

Information Domain Values Simple Average Complex Count FP Count

Number of inputs 4 10 16 10 40
Number of outputs 4 5 16 8 40
Number of inquiries 4 12 19 12 48
Number of files 3 6 10 6 60
Number of external 2 7 3 2 14
interfaces
Count- Total= 202

Complexity weighing factors are determined and the following results are obtained.

Factors (Fi) Value

Backup and recovery 4
Data Communication 1
Distributed processing 0
Performance critical 3
Existing operating environment 2
On-line data entry 5
Input transaction over multiple screens 5
Master file updated online 3
Information domain values complex 3
Internal processing complex 2
Code designed for reuse 0
Conversion / installation in design 1
Multiple installation 3
Application designed for change 5
Sum(Fi) 37

Estimated number of FP:

FPestimated= count-total * [0.65 + .01 * SUM (Fi)]

= 202 * [0.65 + (.01 * 37)]

= 202 * [0.65 + 0.37]
= 202 * 1.02
= 206.04

For historical data, productivity is 55.5 FP / person-month and development cost is $8000 per
month.

Productivity = FP / person-month
FP = 206.04 / 55.5
= 4 person-month

Total cost = Development cost * person-month

= $8000 * 4
= $32000
Cost per FP = Cost / FP
= $32000/206
= $155.34 per FP

Empirical Estimation

Resource models consist of one or more empirically derived equations. These equations are used
to predict effort (in person-month), project duration, or other pertinent project data. There are
four classes of resource models:
 Static single-variable models
 Static multi-variable models
 Dynamic multi-variable models
 Theoretical models
The basic version of the COnstructive COst MOdel or COCOMO is an example of a static
single-variable model. Barry Boehm introduced COCOMO models.
There is a hierarchy of these models.

COCOMO applies to three classes of software projects:

  Organic projects - "small" teams with "good" experience working with "less than rigid"
requirements
 Semi-detached projects - "medium" teams with mixed experience working with a mix of
rigid and less than rigid requirements
 Embedded projects - developed within a set of "tight" constraints. It is also combination
of organic and semi-detached projects.(hardware, software, operational, ...)
The basic COCOMO equationstake the form:

E = ab * (KLOC) bb
db
D = cb * (E)
bb
Effort Applied (E) = ab(KLOC) [ person-months ]
Development Time (D) = cb(Effort Applied)db[months]
People required (P) = Effort Applied / Development Time [count]

Where, KLOC is the estimated number of delivered lines (expressed in thousands) of code for
project. The coefficients ab, bb, cb and db are given in the following table:

Project class ab bb cb db
organic 2.4 1.05 2.5 0.38
semidetached 3.0 1.12 2.5 0.35
embedded 3.6 1.20 2.5 0.3

Example:

Use the information given in the table below

Project Class ab bb cb db
Organic 2.4 1.05 2.5 0.38
Semidetached 3.0 1.12 2.5 0.38
Embedded 3.6 1.20 2.5 0.12

Assume that the expected size of an organic software project is 45.6 KLOC. Using the
coefficients in the above table and the BASIC COCOMO MODEL, calculate the recommended
number of people needs to be engaged to complete the project.

Effort Applied (E) = ab * (KLOC) bb

= 2.4 * (45.6) 1,05
= 132.47
 db
Development Time (D) = cb * (E)
0.38
= 2.5 *(132.47)
= 2.5 * 6.4
= 16
People Required (P) = E/D
= 132.47/16
= 8.27
= 9 people
The Intermediate Cocomo formula takes the form:
E=ai(KLoC)(bi).EAF

Where E is the effort applied in person-months, KLOC is the estimated number of thousands of
delivered lines of code for the project, and The product of all effort multipliers results in an
effort adjustment factor (EAF).The coefficient ai and the exponent bi are given in the next table.
And the ci value is always 2.5.

Software project ai bi
Organic 3.2 1.05
Semi-detached 3.0 1.12
Embedded 2.8 1.20

The Development time D calculation uses E in the same way as in the Basic COCOMO.

Example:
Consider an organ project where three module to implement. User interfaces 0.10 KLOC,
database management 0.8 KLOC and report generation 0.18 KLOC. Efforts and project class
value as follows:

Software
project ai bi di
Class
Organic 3.2 1.05 1.05
Semi-detached 3.0 1.12 1.12
Cost Drivers Level 2.8
Embedded Effort
1.20Adjustment
1.20 Factor
Complexity High 1.15
Storage High 1.06
Experience Low 1.13
Programming Low 1.17
Capabilities
Analyst capability High 1.23
Memory Constraints Low 1.01
Calculate the effort applied in person-month using Intermediate COCOMO Model. What will
be total development time and how many people should be hired?

Solution:

Effort (E) =ai(KLOC)(bi)EAF

= 3.2 (0.10 + 0.8 + 0.18)1.05 * (1.15* 1.06* 1.13* 1.17*1.23*1.01)

= 3.2 * 1.08 1.05 * 2.002

= 6.945

db
Development Time, D = cb * (E)

= 2.5 * 6.9451.05

= 19.129

People, N = E/D = 6.945/19.129 = 0.36