Software Cost Estimation
Approach
I:
Using the COCOMO model
The Basic
COCOMO Model is used, and the type of software development mode is the
semi-detached mode.
The
equations used are listed below.
- E
= ab(KLOC)bb
- D = cb(E)db
where
E is the effort applied in person-months,
D is the development time in chronological months,
KLOC is the estimated number of delivered lines of code for
the project (expressed in thousands).
Table of constants used for calculations
|
Software
Project
|
ab
|
bb
|
cb
|
db
|
|
Organic
Mode
|
2.4
|
1.05
|
2.5
|
0.38
|
|
Semi-detached
Mode
|
3.0
|
1.12
|
2.5
|
0.35
|
|
Embedded
Mode
|
3.6
|
1.20
|
2.5
|
0.32
|
Calculations
The number of lines of code estimated is 2000.
Effort
required, E = (3.0)(2)1.12
= 6.89
person-months
@
7 person-months
Development
Time, D = (2.5)(6.89)0.35
=
4.9 months
Productivity, P = KLOC / D
= 2 / 6.89
= 0.29
Klines of code per person-month
Approach
II:
Using Function Point Analysis
Three
fundamental equations used to compute function points are:
- FP
= UFP x TCF
- TCF
= 0.65 + 0.01 x
DI
- DI
= SFi
where
FP is the Function Points,
UFP (Unadjusted Function Points) is the count-total,
TCF is the Technical Complexity
Factor,
DI is the Degree of Influence,
and Fi are the Factors of influence.
To
calculate DI, the following factors are rated on a scale of 0 to 5:
0
1
2
3
4
5
No
Incidental
Moderate
Average
Significant
Essential
Influence
|
S/N
|
Question
|
Rating
|
|
1
|
Does
the system require reliable backup and recovery?
|
5
|
|
2
|
Are
data communications required?
|
3
|
|
3
|
Are
there distributed processing functions?
|
0
|
|
4
|
Is
performance critical?
|
3
|
|
5
|
Will
the system run in an existing heavily utilized operational environment?
|
0
|
|
6
|
Does
the system require online data entry?
|
5
|
|
7
|
Does
the online data entry require the input transaction to be built over
multiple screens or operations?
|
0
|
|
8
|
Are
the master files updated online?
|
2
|
|
9
|
Are
the inputs, outputs, files, or inquires complex?
|
3
|
|
10
|
Is
the internal processing complex?
|
4
|
|
11
|
Is
the code designed to be reusable?
|
4
|
|
12
|
Are
conversion and installation included in the design?
|
4
|
|
13
|
Is
the system designed for multiple installations in different
organizations?
|
5
|
|
14
|
Is
the application designed to facilitate change and ease of use by the
user?
|
5
|
Total
(DI) 43
TCP = 0.65 + 0.01 x
DI = 1.08
Computation Table
|
Measurement
Parameter
|
Weighting
Factor
|
Sub-total
|
|
Simple
|
Average
|
Complex
|
|
Count
|
Value
|
Count
|
Value
|
Count
|
Value
|
|
Number
of user inputs
|
0
|
3
|
7
|
4
|
0
|
6
|
28
|
|
Number
of user outputs
|
1
|
4
|
0
|
5
|
0
|
7
|
4
|
|
Number
of user inquires
|
1
|
3
|
0
|
4
|
0
|
6
|
3
|
|
Number
of files
|
0
|
7
|
3
|
10
|
0
|
15
|
30
|
|
Number
of external interfaces
|
0
|
5
|
0
|
7
|
0
|
10
|
0
|
|
Count-total
(UFP)
|
|
65
|
|
Complexity
Multiplier (TCF)
|
1.08
|
|
Function
Points (FP)
|
70.2
|
The total
number of function points is 70.2. The value 70.2 signifies the complexity of
the project. However, cost estimates such as effort and development time
cannot be directly calculated from the function points. These values are
obtained by comparison of previously completed projects and their respective
efforts and development times.
For
example, assume that there is a completed project with 80 function points and
effort of 90 person-months. If the current project has 70 function points, by
comparison, the effort for this project would be (80/90) ´
70 = 79.
These
function points can also be used to calculate other attributes for cost
estimation such as software productivity, quality and others, such as cost.
The formulae used are given below.
Productivity
= FP / person-months
Quality =
defects / FP
Cost = $
/ FP
Evaluation
of the two approaches
The
COCOMO model relies heavily on the team’s level of expertise with the
programming language used. Also, it will penalise the implementations with
greater lines of code, which achieves the same effect as one with fewer lines
of codes. Classification of the project and the type of COCOMO model will
indirectly affect the accuracy of the values calculated for the effort,
development and productivity. Hence, there is a greater amount of ambiguity.
The
Function Points approach does not rely on the factors used in the COCOMO
model. It uses the readily countable information of the problem to calculate
the points, thus making the estimation more accurate, and decreasing its
ambiguity. In addition, this approach allows reusability of measures and helps
in object-oriented approach of the software design, which is the approach the
group is taking.
However,
since the group has not worked on other projects previously, it is not
possible to calculate the effort and development time from this approach.
In
conclusion, the group will make use of the both the COCOMO Model approach and
the Function Points approach as a measure of software cost estimation. This
would allow for a compromise between the two methods and giving a more
accurate cost estimate of this project.
Directions
