You are on page 1of 9

Indian Journal of Science and Technology, Vol 7(6), 795803, June 2014

ISSN (Print) : 0974-6846


ISSN (Online) : 0974-5645

A Novel PSO based Approach with Hybrid


of Fuzzy C-Means and Learning Automata
in Software Cost Estimation
Farhad Soleimanian Gharehchopogh*, Laya Ebrahimi, Isa Maleki and Saman Joudati Gourabi
Department of Computer Engineering, Science and Research Branch, Islamic Azad University,
West Azerbaijan, Iran; bonab.farhad@gmail.com, llaya.ebrahimi@gmail.com,
maleki.misa@gmail.com, saman.jodati@gmail.com

Abstract
We used Particle Swarm Optimization (PSO) algorithm hybrid with Fuzzy C-Means (FCM) and Learning Automata (LA)
algorithms for Software Cost Estimation (SCE). In this paper we test and evaluate PSO-FCM and PSO-LA hybrid models on
NASA dataset software projects. The obtained results showed that in the hybrid models the values of Magnitude of Relative
Error (MRE) and Mean Magnitude of Relative Error (MMRE) were reduced compared with COCOMO model and also the
accuracy of Percentage of Relative Error Deviation (PRED) was higher in the hybrid models.

Keywords: COCOMO Model, Fuzzy C-Means, Learning Automata, Particle Swarm Optimization, Software Cost Estimation

1.Introduction

Software Cost Estimation (SCE) is an important issue in


software project management. The costs increase with
increasing the complexity of the software. Therefore the
accuracy of the estimation in early stages of the project is
extremely favorable. In fact, the cost and timely finishing
of software projects are among decisive factors of prosperity of a software development company. Other strategic
issues such as timing and experienced manpower are also
effective in the success of the project.
SCE is a sensitive, complicated and inevitable issue in
software development. In the last three decades a dramatic
growth has been observed in using different cost estimation models15. Along this growing process, it is noticeable
that in fact in all SCE models, the main goal is the easy
and fast development, planning and scheduling of software projects. Despite the fact that it cannot be proved
that SCE algorithm models have precise performance in
cost and manpower, a precise estimation is considered a
competitive advantage for software companies and guides
*Author for correspondence

the companies in better analyzing the feasibility investigation and efficient management of the projects6,7. The
main reasons in unsuccessful conclusion of the projects
according to the cost and schedule are failure to use management methodology, the quality of tool application and
imprecise estimation. The topic of SCE in software project management that include are planning for resources,
costs, employees and time control, has become a major
concern for software companies8,9.
Among SCE models, COCOMO I10, COCOMO II11,
SLIM12, and Function Point (FP)13 models are in algorithmic models category. The project managers cannot make
precise and reliable estimations on final status of the projects in respect with the required time and cost to finish
the project relying on COCOMO model. In algorithmic
models, estimation parameters are often obtained from
experimental data of different and previous projects14.
SCE algorithmic models work essentially based on cost
factors and scale factors. Also the efficiency of these models depends on the size of the projects and variation in
project size results in numerous subsequent variations

A Novel PSO based Approach with Hybrid of Fuzzy C-Means and Learning Automata in Software Cost Estimation

in the cost value. The wrong values of cost factors may


result in major variations in the final results of SCE.
Therefore in this paper we aim to evaluate the effective
factors of estimation and make more precise project cost
estimations using hybrid PSO algorithm with FCM and
LA algorithms. PSO algorithm for solving combinatorial
optimization problems is to have good performance1517.
Software production and development is among major
investments of the software companies and software engineers are often worried about the precise estimation and
quality of the software projects. Therefore, with adequate
knowledge about SCE, a software project manager can
control and conclude development of software projects in
precise time. Exact evaluation of SCE in project planning
and conducting is considered among important project
management activities. The related investigations show
that the techniques used in estimation often work better
than algorithmic models.
A case study based on ANNs for SCE is presented1.
The results showed that Artificial Neural Networks
(ANNs) model is one of the best models for SCE. The
training results of ANNs show that in more than 90% of
the cases they make better estimations. The investigations
resulted from ANNs tests showed that using artificial
intelligence techniques have much better performance
compared to algorithmic models and make more precise
estimations. Despite the approximate value and error estimation in both methods, it should be kept in mind that in
all by using artificial intelligence methods, the error value
would be very small.
Soft computing techniques have been used in order
to deal with ambiguity and effective factors in SCE18. The
main goal in this study is investigating the role of fuzzy
logic in improving the accuracy of estimation model
COCOMO II through determining input parameters
using Gaussian, triangular and trapezoidal membership
functions. NASA (93) dataset was used in the evaluation of COCOMO II model. The results show that SCE
gets significantly improved compared with COCOMO II
model by applying Gaussian membership function and
fuzziness scale factors. And by fuzziness the MRE error
value is significantly reduced.
The researchers18 presented a new model for SCE
based on Multi-Objective Particle Swarm Optimization
(MOPSO) algorithm. In order to test and evaluate
MOPSO model the COCOMO dataset was used. The
test results showed that MOPSO model had better performance compared with COCOMO the value of mean

796

Vol 7 (6) | June 2014 | www.indjst.org

absolute relative error was reduced. Investigation and


evaluation of SCE was achieved with Genetic Algorithm
(GA) combination and Ant Colony Optimization
(ACO)5. In the hybrid model effective factors in estimation were tested by GA and trained by ACO and better
results were achieved compared to COCOMO. Their test
results showed that the combination of GA and ACO had
better performance in software project cost estimation
compared to COCOMO and there was also smaller MRE.
This paper is organized as follows: in Section 2, we
introduce meta-heuristic algorithms; in Section 3, we
describe the proposed models; in Section 4, the evaluation and results of the proposed models are presented and
finally in Section 5, the conclusion and future works are
presented.

2. Meta-Heuristic Algorithms
The ability of meta-heuristic algorithms in solving optimization problems works through factors collective
cooperation and parallel search. Therefore the more the
power of the meta-heuristic algorithms in controlling
these two factors, the more the ability of these algorithms
in finding solutions close to the optimal solution. Metaheuristic algorithms based on collective intelligence deal
with optimal problem solvation, are tolls in finding close
to optimal solutions. In this section we introduce PSO
algorithm which is one of the most important population
algorithms.

2.1 Particle Swarm Optimization


PSO algorithm was introduced in 1995 inspired by
the social behavior of birds that live in small and large
groups20. PSO is a simulation of group social behavior
of birds that search for food in an environment. None
of the birds has information about the position of food
but in each stage they know how far they are from the
food. Based on this, the best way to find food is following the bird nearest to the food. The PSO algorithm is a
population algorithm in which a few particles that are the
solutions of a function or problem form a population. A
population of particles moves in the problem space and
based on their personal experience try to find the optimal
solution in the search space. PSO algorithm as an optimization algorithm forms a population based search in
which each particle changes its position in time. In PSO
algorithm, the particles in move in a multi-dimensional

Indian Journal of Science and Technology

Farhad Soleimanian Gharehchopogh, Laya Ebrahimi, Isa Maleki and Saman Joudati Gourabi

search space of possible solutions. In this space an evaluation criterion is defined and the quality assessment of the
problem solutions is achieved according to that. The variation of status of each particle in a group is influenced by
its own experience and its neighbors knowledge and the
search behavior of one particle affects the other particles
in the group. This simple behavior results in formation of
optimal areas in search space. Therefore in PSO algorithm
each particle informs other particles in a suitable way once
it finds the optimal position and based on the achieved
values for cost function, each particle decides according
to a certain possibility, to follow other particles and the
search in problem space is done based on the previous
knowledge of the particles. This results in the particles not
getting too close to one another and helps them efficiently
solve the continuous optimization problems.
In PSO algorithm first the individuals in the groups
are accidentally created in the problem space and the
search for the optimal answer begins. In the overall structure of the search each person follows the individual with
the best fitness function, while not forgetting his own
experience and follows the situation in which he has the
best fitness function itself. Therefore each person in each
algorithm iteration changes his future position according to two values, one being the best personal position
so far (pbest) and the other the best position of all population so far and in fact the best pbest in the whole
population (gbest). Conceptually, pbest for each person is
in fact the biological memory of that person. Gbest is the
general knowledge of the population and when people
change their position according to gbest in fact they try
to upgrade his own knowledge to the level of population
knowledge. Conceptually the best particle group connects
every particle in the group to one another. Determination
of the next position of the particle for each particle is
achieved using Equations (1) and (2).
vi +1 = w.vi + c1 .r1 .(Pbesti xi ) + c2 .r2 .( g besti xi ),

(1)

xi +1 = xi + vi +1

(2)

In Eq. (1) c1 and c2 are learning parameters. Rand () is


a function to produce random numbers in the range [0,
1]. Xi is the current position and vi is the velocity of the
people. W is a control parameter which controls the current velocity (vi) with the effect of the next velocity and
creates a balance between the ability of the algorithm in
local and global search and therefore reaches the answer
in a shorter mean time. Therefore in order to achieve

Vol 7 (6) | June 2014 | www.indjst.org

optimal performance of the algorithm, parameter w is


defined by Eq. (3):
w = w Max

((w Max w Min ) i)


iMax

(3)

In Eq. (3) imax shows the maximum number of algorithm iteration and i is the counter of the optimal answer
iteration. In Eq. (3) parameters wmax and wmin are the initial and ultimate values of inertia weight along algorithm
implementation. The value of inertia weight changes
linearly from 0.9 to 0.4 along the implementation of the
algorithm. The high values w result in general search and
smaller values result in local search. In order to make
a balance between local and general searches, the inertia weight along the implementation of the algorithm
is reduced uniformly. Therefore with decreasing the W
value the search is more local and around the optimal
answer.

3. Proposed Models
Precise and reliable estimation for software projects is
among the current challenges in software engineering.
The accuracy of cost estimation is one of the major phases
in interactions between manpower and time. Precise
estimation is a complicated process since the predicted
results should be close to reality. SCE algorithmic models
work essentially based on cost and scale factors. Therefore
the value of these factors in sensitivity of cost and manpower is significantly effective. Also the majority of these
models depend on the project size and the variation in the
factors of the projects size result in numerous variations
in cost and time. Wrong calculations and predictions of
cost factors will also have major effects in the final result.
Determination of cost factor values in software projects
is very difficult and complicated. In hybrid models Effort
Multipliers (EM) factors19 such as RELY, CPLX, STOR,
TOOL and SCED which are very effective in estimation
accuracy are evaluated and tested.
FCM clustering algorithm is the most widely used
algorithm in recognizing interrelated data in different
clusters22. In FCM algorithm, at first random points are
selected equal in number to the required clusters and then
the data are assigned to one of these clusters according to
the proximity and the clusters are formed. By repeating
this procedure new centers can be calculated for the data
through calculating the mean in each iteration and then
the data are reassigned to new clusters. This procedure

Indian Journal of Science and Technology

797

A Novel PSO based Approach with Hybrid of Fuzzy C-Means and Learning Automata in Software Cost Estimation

is repeated until no variation in the data is observed. In


FCM clustering algorithm there is an objective function
which defines the distance between the data. The objective function in this algorithm is defined as Eq. (4):
c

Jm =

m
ij

u
i =1 j 1

x j vi

(4)

Of course the following condition should be valid.


c

u
i =1

ij

j = 1,..., n

=1

(5)

In Eq. (4), c is the cluster number, n is the data number, m is the fuzziness amount which can be a real number
bigger than 1. Xj is the kth data, vi is the ith cluster and uij
is the membership degree of jth data in ith cluster. In all
fuzzy clustering algorithms at first the cluster number is
defined and the primary values are assigned to the clusters. Then using Equations (6) and (7), these values are
updated and this procedure is repeated until a certain difference between the data is achieved.
n

(u )
vi =

j =1
n

ij

determine the cost factors. Figure 1 shows the flowchart


for the PSO-FCM hybrid model.
Figure 2 shows the quasi code for the PSO-FCM
hybrid model.
The LA is a machine with limited states that can do
limited functions20. Each selected function is evaluated with possible environment and gives an answer to
the learning automata. The learning automata use this
answer and choose the nest function. In this process,
the LA learns to how to choose the best function. The
LA with changing structure is showed with {, , p, T}

xj
(6)

(u )
ij

j =1

uij =

1
c

x j vi

x
i =1

2 /( m 1)

(7)

vi

Generally the FCM clustering is used to find the structure in the data which are not labeled. In this situation it
is tried to form an objective function which has a minimum by putting the data in different clusters. Therefore
by selecting fuzzy distance functions the data can be optimally clustered.
In hybrid PSO-FCM model the minimum distance
between the clusters and distance summation inside the
cluster and the number of the clusters are used as fitness parameters and to improve the PSO algorithm.
Application of FCM results in the accumulation of the
data in the best cluster and the fitness function to have
many local optimal points. Each particle in the group
possesses a set of FCM rules. In different generations
the particles get different values which improve the fitness function. Ultimately the cluster with the highest
fitness of the particles is selected as the best set of rules to

798

Vol 7 (6) | June 2014 | www.indjst.org

Figure 1. Flowchart PSO-FCM hybrid model.

Figure 2. PSO-FCM Hybrid Model Quasi Code.

Indian Journal of Science and Technology

Farhad Soleimanian Gharehchopogh, Laya Ebrahimi, Isa Maleki and Saman Joudati Gourabi

in which is the set of automata functions, is the set


of automata inputs, p is the selection probability vector
of each factor and p(n+1)=T[(n), (n), p(n)]. The LA
algorithm is defined linearly by Equations (8) and (9)
shows the update quality of LA probability vector when
the automata receive favorable answer from the environment. Eq. (8) shows the update quality of LA probability
vector when the automata receive favorable answer from
the environment.
pi (n + 1) = pi (n) + a[1 pi (n)] if i = j
(8)
p j (n + 1) = (1 ). p j (n) if i j
Eq. (9) shows the update quality of LA probability
vector when the automata receive unfavorable answer
from the environment.
pi (n + 1) = (1 b). pi (n)

if i = j

p j (n + 1) = (b / r 1) + (1 b) p j (n)

if i j

(9)

In Equations (8) and (9) parameters a and b are the


amount of award and punishment respectively.
In order to improve the performance of PSO algorithm LA used to adjust the behavior of the particles. In
hybrid PSO-LA model all particles make local search in
the search space. In hybrid PSO-LA model the LA model
strategy, according to the award criterion for PSO algorithm makes it possible for the particle to achieve many
local bests. In the hybrid model, particles share their fitness in a similar area and these results in competition of
the people in low density population areas with the people in high density population areas and distribute in the
environment. In this state a suitable diversity with particles with better fitness is created which can find local
bests. Therefore, in hybrid PSO-LA model if particles in
one area had better group experiences compared with
particles in other areas, they are awarded with LA model
and these results in maintaining the diversity of the particles among the particles and the velocity and position
vectors of the particles are updated in order not to put the
particles in premature convergence. Figure 3 shows the
flowchart of the PSO-LA hybrid model.
Figure 4 shows the quasi code of the PSO-LA hybrid
model.
In the hybrid models, MMRE is considered as the
objective function. The aim of fitness function in the recommended models is minimizing the value of MMRE
compared with COCOMO algorithm model and the

Vol 7 (6) | June 2014 | www.indjst.org

Figure 3. Flowchart PSO-LA Hybrid Model.

Figure 4. PSO-LA Hybrid Model Quasi Code.

hybrid models are iterated until the MMRE value is


reduced to a desirable amount. MMRE is defined with
Eq. (11)21.
MREi =

Actuali Estimatei

MMRE =

Actuali
1
N

MRE
i =1

100

(10)
(11)

Using Eq. (11) the error summation between hybrid


models and COCOMO model can be compared, also the
Percentage of Relative Error Deviation (PRED) which
is an important criterion is considered for the accuracy

Indian Journal of Science and Technology

799

A Novel PSO based Approach with Hybrid of Fuzzy C-Means and Learning Automata in Software Cost Estimation

of the prediction. The most commonly used prediction


methods are MMRE and PRED. PRED(x) is defined with
Eq. (12)24:
n
1, if MRE x
1
PRED(x ) =
n i =1 0 , otherwise

(12)

The criterion PRED(x) which is defined based on


MRE is the most commonly used method in accuracy
prediction and has a good expression of how the recommended model works.

4. Evaluation and Results


In this section the results of hybrid model simulation for
60 projects from NASA software projects21 are studied.
The simulation of hybrid models was conducted by using
C#.NET 2008 programming language. In PSO algorithm,
in order to make balance and search in different areas, the
values of algorithm parameters need to be precisely determined. Two key factors that have direct influence on the
efficiency PSO algorithm are algorithm coefficients and
population size. The algorithm coefficients are a set of real
numbers which determine the procedure of optimization
and particle trajectory in the search space. Table 1 shows
the parameters used in hybrid models.
In Table 1 parameter w is used to control the effects of
particle velocity in the previous moment on the velocity
at the current moment. Parameters c1 and c2 determine
the effect of particle learnings on optimal search.
Parameter c1 determines to some extend the particles
are relying on their own experience and parameter c2
determines the amount the particles rely on their neighbors experiences. In PSO-FCM hybrid model because

the objective is clustering and non-clustering of particles


in the search space, the number of clusters are assumed
to be equal to 2. Parameter m in the hybrid PSO-FCM
model shows the amount of fuzziness. The values of the
parameters a and b are the award and punishment used
for the particles respectively. In PSO algorithm which
is based on random search, cannot results on one iteration of the software regarding the convergence velocity
and achieving answer. The reason for this is existence
of randomness and probability in the structure of PSO
algorithm. Therefore the program runs 100 times in
average.
Figure 5 shows the MRE comparison between PSOFCM hybrid model and intermediate COCOMO among
60 projects from NASA dataset using clustering. As it can
be seen the PSO-FCM hybrid model has smaller error
rate compared to intermediate COCOMO model.
PSO-FCM hybrid model in order to reach the optimal
composition and their clustering and more precise evaluation of MRE criterion, the numbers of the clusters were
assumed to be 4. As in Figure 6, the results of PSO-FCM
hybrid model which are shown in red have more tendencies to the inside compared to COCOMO model and this
shows that their error is smaller. According to the results
obtained from the variation of cluster number it can be
said that the performance results of the hybrid algorithm
is improved.
In Table 2 we have evaluated and compared 10 projects from NASA dataset software projects. The result of
Table 2 shows that the PSO-FCM hybrid model has lowered MRE compared to COCOMO model. Therefore,
the PSO-FCM hybrid model was useful for the estimation and the estimation error was lower compared to
COCOMO model.

Table 1. Parameter Values

800

Parameters

Value

Population Size

50

WMAX

0.9

WMIN

0.4

C1

1.5

C2

1.5

Iterations

100

1.25

1.05

1.05

Vol 7 (6) | June 2014 | www.indjst.org

Figure 5. MRE Comparison between PSO-FCM Hybrid


Model and Intermediate COCOMO Model.

Indian Journal of Science and Technology

Farhad Soleimanian Gharehchopogh, Laya Ebrahimi, Isa Maleki and Saman Joudati Gourabi

In Table 3, we have evaluated and compared 10 projects from NASA dataset software projects. As can be seen
the PSO-LA hybrid model has lower error rate compared
to intermediate COCOMO model.
MRE comparison curve between PSO-LA hybrid
model and intermediate COCOMO model on 10 projects from NASA dataset software projects are shown in
Figure8.
Figure 9 shows MRE comparison between PSO-FCM
and PSO-LA hybrid model and intermediate COCOMO
model on 10 projects from NASA dataset software projects. As can be seen the PSO-FCM hybrid model has lower
error rate compared to PSO-LA hybrid model. Also the
MRE error rate in PSO-FCM and PSO-LA hybrid models
is lower compared to intermediate COCOMO model.
Table 4 shows MMRE comparison between hybrid
models and intermediate COCOMO model for 60 projects from NASA dataset software projects. The results

Figure 6. MRE Comparison between PSO-FCM Hybrid


Model and Intermediate COCOMO Model.

Table 2. MRE Comparison between PSO-FCM


Hybrid Model and COCOMO Model on 10 Projects
No.

Project
No.

Models
COCOMO

PSO-FCM
No. Clusters
1

Figure 7. MRE Comparison between PSO-FCM Hybrid


Model and Intermediate COCOMO Model on 10 Projects.

11

27.21

2.20

3.42

1.86

2.15

15

9.36

11.20

8.56

13.15

7.91

19

3.8

5.23

6.12

7.32

3.5

21

22.51

10.78

13.98

11.2

9.61

Table 3. MRE Comparison between


PSO-LA Hybrid Model and Intermediate
COCOMO Model

25

39.54

12.85

10.84

16.87

9.62

No

30

31.56

12.79

17.15

15.48

9.31

35

28.46

12.76

8.65

11.35

2.76

39

23.76

15.12

12.34

5.23

44

24.50

32.41

26.14

10

55

3.64

8.16

9.56

6.31

Project
No.

COCOMO

PSO-LA

11

27.21

4.56

2.72

15

9.36

11.23

21.48 17.45

19

3.8

6.72

21

22.51

8.69

25

39.54

13.25

30

31.56

17.61

35

28.46

7.41

39

23.76

16.35

44

24.50

12.97

10

55

3.64

8.54

4.56

Figure 7 shows the MRE comparison between PSOFCM hybrid model and intermediate COCOMO model
on 10 projects from NASA dataset software projects.
As can be seen in the clustering of PSO-FCM hybrid
model the error rate is smaller compared to intermediate
COCOMO model.

Vol 7 (6) | June 2014 | www.indjst.org

Models

Indian Journal of Science and Technology

801

A Novel PSO based Approach with Hybrid of Fuzzy C-Means and Learning Automata in Software Cost Estimation

Table 5. PRED Comparison between Hybrid


Models and COCOMO Model
Models
PRED(X)

COCOMO

PSO-FCM

PSOLA

Cluster
1

Figure 8. MRE Comparison between PSO-LA Hybrid


Model and Intermediate COCOMO Model on 10 Projects.

Figure 9. MRE Comparison between Hybrid Models and


Intermediate COCOMO Model on 10 Projects.

Table 4. MMRE Comparison


between Hybrid Models and
COCOMO Model.
Models

MMRE

COCOMO
No. Clusters

PSO-FCM

29.6

PSO-LA

25.36

24.56

24.22

23.86

26.32

of Table 4 shows that the hybrid models have reduced


MMRE error value and it can be said that these are helpful methods for SCE.
As can be seen in Table 4 the hybrid models have lower
MMRE error compared to COCOMO model. Table 5
compares PRED between hybrid models and intermediate COCOMO model for 60 projects from NASA dataset
software projects.
As can be seen in Table 5 in all cases the hybrid models had higher PRED accuracy compared to COCOMO

802

Vol 7 (6) | June 2014 | www.indjst.org

65

68.3

63.3

43.3 38.3 41.6

50

36.3

16.6

31.6 23.3 26.6

35

23.3

11.6

18.3

13.3

11.6

PRED(25)

40

PRED(15)

18.3

PRED(10)
PRED(5)

61.6 58.3

8.3

model. Among the hybrid models PSO-FCM model had


higher PRED accuracy compared to PSO-LA model.
Therefore it can be said that the hybrid models may also
have some advantages over one another and their convergence may have a great effect in accuracy and prediction.
In order to evaluate the hybrid models three types of comparison namely MRE, MMRE and PRED was done from
the accuracy viewpoint. After investigating the results
of the hybrid modes it was shown that the probability
of convergence achieving answer with lower error was
higher in PSO-FCM hybrid model compared to PSO-LA
hybrid model.

5. Conclusion and Future Works


The SCE is one of the most sensitive, complicated and
inevitable issues in the process of developing software. In
the last three decades a significant growth was observed
in using different types of SCE. In this increasing process
it is possible that in fact in all cost estimation models, the
major objective of software development cost reduction
and easy and fast scheduling. The inherent ambiguity of
the software development project results in the fact that
there is a not high expectation for precise estimation.
Algorithmic models in the field of estimation and simulation need numerous input parameters whose precise
determination is difficult and their measurement also
needs a lot of time and cost. In this article the hybrid
models for SCE using PSO algorithm are presented. The
test results have showed that the PSO-FCM and PSO-LA
hybrid models have higher performance compared to
COCOMO model and achieve higher accuracies in SCE.
Because of approximate values and error estimations in
hybrid models it should be kept in mind that in all using
artificial intelligence can reduce error rate. According
to the results of this paper it could be said that hybrid

Indian Journal of Science and Technology

Farhad Soleimanian Gharehchopogh, Laya Ebrahimi, Isa Maleki and Saman Joudati Gourabi

algorithms based on PSO can significantly improve the


accuracy of the management decision making in the
development of software projects. We hope that in future
obtain better results through combining meta-heuristic
algorithms.

6.References
1. Gharehchopogh FS. Neural networks application in software cost estimation: a case study. 2011 International
Symposium on Innovations in Intelligent Systems and
Applications (INISTA 2011). 2011 1518 Jun; Istanbul,
Turkey: IEEE. p. 6973.
2. Khalifelu ZA, Gharehchopogh FS. Comparison and evaluation data mining techniques with algorithmic models in
software cost estimation. Procedia-Technology Journal.
2012; 1:6571.
3. 
Khalifelu ZA, Gharehchopogh FS. A new approach in
software cost estimation using regression based classifier.
AWERProcedia Information Technology & Computer
Science Journal. 2012; 2. 25256.
4. Khalifelu ZA, Gharehchopogh FS. A survey of data mining techniques in software cost estimation. AWERProcedia
Information Technology & Computer Science Journal.
2012; 1:33142.
5. Maleki I, Ghaffari A, Masdari M. A new approach for software cost estimation with hybrid genetic algorithm and ant
colony optimization. International Journal of Innovation
and Applied Studies. 2014; 5(1):7281.
6. Li YF, Xie M, Goh TN. A study of project selection and feature weighting for analogy based software cost estimation. J
Syst Software. 2009; 82(2):24152.
7. Mittas N., Angelis L. Visual comparison of software cost
estimation models by regression error characteristic analysis. J Syst Software. 2010; 83(4):62137.
8. Jorgensen M, Shepperd M. A systematic review of software
development cost estimation studies. IEEE Trans Software
Eng. 2007; 33(1):4053.
9. Anish M, Kamal P, Harish M. Software cost estimation
using fuzzy logic. ACM SIGSOFT Software Engineering
Notes. 2010; 35(1):115.
10. Boehm BW. Software Engineering Economics. Englewood
Cliffs, New Jersy: Prentice-Hall; 1981.

Vol 7 (6) | June 2014 | www.indjst.org

11. Boehm BW. Software Cost Estimation with COCOMO II.


PTR, Englewood Cliffs, New Jersy: Prentice Hall; 2000.
12. Putnam LH. A general empirical solution to the macro software sizing and estimation problem. IEEE Trans Software
Eng. 1978; 4(4):34561.
13. Albrecht AJ, Gaffney JE. Software function, source lines
of code, and development effort prediction: a software
science validation. IEEE Trans Software Eng. 1983 Nov;
9(6):63948.
14. 
Cuadrado-Gallego JJ, Fernandez-Sanz L, Sicilia M-A.
Enhancing input value selection in parametric software
cost estimation models through second level cost drivers.
Software Qual J. 2006; 14(4):33957.
15. Maleki I, Khaze SR, Tabrizi MM, Bagherinia A. A new
approach for area coverage problem in wireless sensor
network with hybrid particle swarm optimization and differential evolution algorithms. International Journal of
Mobile Network Communications & Telematics. 2013 Dec;
3(6):6175.
16. Gharehchopogh FS, Maleki I, Zebardast B. A new solutions
for continuous optimization functions by using bacterial
foraging optimization and particle swarm optimization
algorithms. Elixir Comp. Sci. & Engg. 2013 Jul; 61:16655
16661.
17. Gharehchopogh FS, Maleki I, Khaze SR. A new optimization method for dynamic travelling salesman problem
with hybrid ant colony optimization algorithm and particle
swarm optimization. International Journal of Advanced
Research in Computer Engineering & Technology. 2013
Feb; 2(2):35258.
18. Reddy PVGDP, Hari CHVMK, Rao TS. Multi objective
particle swarm optimization for software cost estimation.
Int J Comput Appl. 2011; 32(3):1317.
19. Menzies T, Port D, Chen Zh, Hihn J. Validation Methods
for Calibrating Software Effort Models. ICSE ACM, USA;
2005.
20. Narendra KS, Thathachar MAL. Learning Automata: an
Introduction. Prentice-Hall Inc; 1989.
21. 
MacDonell SG, Gray AR. A comparison of modeling techniques for software development effort
prediction. Proceedings of International Conference on
Neural Information Processing and Intelligent Information
Systems; 1997. p. 86972.

Indian Journal of Science and Technology

803

You might also like