International Journal of Trend in Scientific Research and Development (IJTSRD)

Volume 3 Issue 5, August 2019 Available Online: www.ijtsrd.com e-ISSN: 2456 – 6470

Review on Algorithmic and Non-Algorithmic

Software Cost Estimation Techniques
Pa Pa Win, War War Myint, Hlaing Phyu Phyu Mon, Seint Wint Thu
Faculty of Information Science, University of Computer Studies, Meiktila, Myanmar

How to cite this paper: Pa Pa Win | War ABSTRACT

War Myint | Hlaing Phyu Phyu Mon | Seint Effective software cost estimation is the most challenging and important
Wint Thu "Review on Algorithmic and activities in software development. Developers want a simple and accurate
Non-Algorithmic Software Cost method of efforts estimation. Estimation of the cost before starting of work is a
Estimation Techniques" Published in prediction and prediction always not accurate. Software effort estimation is a
International very critical task in the software engineering and to control quality and
Journal of Trend in efficiency a suitable estimation technique is crucial. This paper gives a review
Scientific Research of various available software effort estimation methods, mainly focus on the
and Development algorithmic model and non – algorithmic model. These existing methods for
(ijtsrd), ISSN: 2456- software cost estimation are illustrated and their aspect will be discussed. No
6470, Volume-3 | single technique is best for all situations, and thus a careful comparison of the
Issue-5, August IJTSRD26511 results of several approaches is most likely to produce realistic estimation.
2019, pp.890-895, This paper provides a detailed overview of existing software cost estimation
https://doi.org/10.31142/ijtsrd26511 models and techniques. This paper presents the strength and weakness of
various cost estimation methods. This paper focuses on some of the relevant
Copyright © 2019 by author(s) and reasons that cause inaccurate estimation.
International Journal of Trend in Scientific
Research and Development Journal. This KEYWORDS: Software cost estimation; Software Effort Estimation; algorithmic;
is an Open Access article distributed non-algorithmic
under the terms of 1. INTRODUCTION
the Creative Software cost estimation plays an important role in software engineering, often
Commons Attribution determining the success or failure of contract negotiation and project execution.
License (CC BY 4.0) The main goal of software cost and effort estimation is to scientifically estimate
(http://creativecommons.org/licenses/by the required workload and its corresponding costs in the life cycle of the
/4.0) software system.
Accurate cost estimates activity is critical to developers and
Lack of trained estimators.
customers. Accurate cost estimation is important for the
Software is intangible, invisible, and intractable so it is
following reasons [1][2]: more difficult to understand and estimate a product or

It can be used to classify and prioritize development process that cannot be seen and touched.
projects with respect to the complete business plan.

It can help to find out what resources to commit to the 2. BACKGROUND
project and how well these resources will be used Software project failure has been an important subject in the

It can help to assess the impact of changes and last decade. Software projects usually don’t fail during the
supporting for preplanning. Projects can be easier to implementation and the most project fails during the
manage and control when resources are matched to real implementation and most project fails are related to the
needs. planning and estimation steps despite going to overtime and

Customers expect accurate development costs to be in cost, approximately between 30% and 40% of the software
line with estimated costs. projects are completed and the other fail(Molokken and
Jorgenson,2003). During the last decade, several studies
Software cost estimation activity historically has been a
have been done in term of finding the reason for the
major difficulty in software development. Several reasons
software project failure. Galorath and Evans(2006)
have been identified that affects the cost estimation process
performed an intensive search between 2100 internet site
such as [15]
and found 5500 reasons for software project failures. Among

Cost of software development estimate is difficult. The
the found reasons, insufficient requirements engineering,
first steps in the estimate are to understand and define
poor planning the projects, suddenly decision at the early
the system to be estimated.
stages of the project and inaccurate estimations were the

A cost estimate done early in the project life cycle is
most important reasons. The other researches regarding the
generally based on less precise inputs and less detailed
reason of project fail to show that inaccurate estimation is
design specifications.
the root factor of fail in the most software project

Software development involves many interrelated
fails(Jones,2007; Jorgensen, 2003 ). Despite the indicated
factors, which affect development effort and
statistics may be pessimistic, inaccurate estimation is a real
productivity, and whose relationships are not well
problem in the software product’s world which should be
understood.
solved. Presenting the efficient techniques and reliable

Incomplete, inaccurate or inconsistent historical
models seems required regarding the mentioned problems.
database of cost measurement.

@ IJTSRD | Unique Paper ID – IJTSRD26511 | Volume – 3 | Issue – 5 | July - August 2019 Page 890
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
The conditions of the software projects are not stable and problems, where decision making is very difficult and
the state is continuously changing so several methods should conditions are vague, fuzzy systems are an efficient tool in
be presented for estimation that each method is appropriate such situations. Fuzzy technique always supports the facts
for a special project[9]. that may be ignored. Following four stages in the fuzzy
approach:
3. ESTIMATION TECHNIQUES Stage 1: produce trapezoidal numbers for the linguistic
Generally, there are many methods for software cost terms.
estimation, which are divided into four categories: Stage 2: develop the complexity matrix by producing a new
Algorithmic, Non-Algorithmic, Parametric and Machine linguistic term.
learning Models. All categories are required for performing Stage 3: determine the productivity rate and the attempt for
an accurate estimation. If the requirements are known the new linguistic terms.
better, their performance will be better. In this section Stage 4: determine the effort required to complete a task and
overview of four estimation models are discussed. to compare the existing method. For example in [3] COCOMO
A. Algorithmic Models These models work based on the technique has been implemented by using the fuzzy method.
special algorithm. These model usually need data at first
and make result by using the mathematical relation. Fig (2) displays all following mentioned steps.
Nowadays, many software estimation methods use these Step (1) fuzzification has been done by scale factors, cost
models. Algorithm models are classified into some drivers and size.
different models. Each algorithmic model uses an Step (2) principals of COCOMO are considered.
equation to do the estimation: Effort=f(x1,x2,……….,xn) Step (3) defuzzification is accomplished to find the effort [6].
where (x1……xn) is the vector of the cost factor. The
differences among the existing algorithmic methods are 4. The parametric-estimating method is a mathematical
related to choosing the cost factor and function. representation of cost estimating relationships that provide
a logical and predictable correlation between the cost as a
B. Non-Algorithmic Model Contrary to the Algorithmic dependent variable and the cost estimating factors as the
methods, methods of this group are based on analytical independent variables associated with the project being
comparisons and inferences. For using the Non estimated (Duverlie and Clastelain, 1999; Dysert, 2003;
Algorithmic methods some information about the International Society of Parametric Analysis [ISPA], 2008).
previous projects which are similar to the Parametric models are developed by applying regression
underestimate project is required and usually analysis to historical project data (obtained from past
estimation process in these methods is done according projects).
to the analysis of the previous datasets. Here, three
methods have been selected for accessing because these 4. ALGORITHMIC METHODS
methods are more popular than the other None These methods are designed to provide some mathematical
Algorithmic methods and many papers about their uses equations to perform software cost estimation. These
have been published in recent years. mathematical equations are based on research and historical
data and use some inputs for example Source Lines of Code,
C. MACHINE LEARNING METHODS Most techniques about a number of functions to perform, and some cost drivers like
cost estimation use statistical methods, which are not as language, design methodology, skill-levels, risk
able to present reason and strong results. This approach assessments, etc. Algorithmic methods developed many
could be appropriate because they can increase the models such as COCOMO models, Putnam model, and
accuracy of estimation by training rules of estimation function points based models [3].
and repeating the run cycles. It is categorized into two
main methods, neural networks and fuzzy methods A. COCOMO Model
which are : One very widely used algorithmic cost estimation model is
the Constructive Cost Model (COCOMO) which was proposed
Neural networks include several layers which each layer is by Boehm [4]. The basic COCOMO model has a simple form:
composed of several elements called neuron. Neurons, by MAN-MONTHS = K1* (KDLOC) K2 Where K1 and K2 are two
investigating the weights defined for inputs, produce the parameters which are dependent on the application and
outputs. Outputs will be the actual effort, which is the main development environment. Estimates from the basic
goal of estimation. Backpropagation neural network is the COCOMO model can be made more accurate by taking into
best selection for software estimation problem because it account other factors concerning the required
adjusts the weights by comparing the network outputs and characteristics of the software to be developed, the
actual results. In addition, training is done effectively. qualification and experience of the development team, and
Majority of researches on using the neural networks for the software development environment. The complexity of
software cost estimation are focused on modeling the the software has the following factor:
Cocomo method, for example in [5] a neural network has Reliability
been proposed for estimation of software cost according to
Database size
the following figure. Figure (1) shows the layers, inputs and
Required efficiency for memory and execution time
the transfer function of the mentioned neural network. Scale
The capability of an analyst and programmer
Factors(SF) and effort multipliers(EM) are an input of the
Team experience in the application area
neural network, pi and qj are respectively the weight of SFs
Experience of the team in the programming language
and EMs.[6] and computer

Use of software engineering and tools
Fuzzy Method The systems, which work based on the fuzzy
logic try to simulate human behavior and reasoning. In many

@ IJTSRD | Unique Paper ID – IJTSRD26511 | Volume – 3 | Issue – 5 | July - August 2019 Page 891
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
Cost models generally use some cost indicator for estimation the software equation: Technical constant C= size * B1/3 *
and notice to all specification artifacts and activities. T4/3 Total PM B=1/T4 *(size/C)3 T = Required
COCOMO 81(constructive cost model) proposed by Barry Development Time in years Size = estimated in LOC Where: C
Bohem is the most popular method which categorized in = parameter dependent on the development environment
algorithmic methods. This method uses some equations and and is determined on the basis of historical data of the past
parameters, which have been derived from previous projects. Rating for C=2,000 is poor C=8000 is good
experiences about software projects for estimation. There C=12,000 is excellent. This model is very sensitive to the
are three forms of the constructive cost model: 1. Basic development time, decreasing the development time can
COCOMO which gives an initial rough estimate of man- greatly increase the person-months needed for development
months and development time. 2. Intermediate COCOMO [6][12]. One significant problem with this model is that it is
which gives a more detailed estimate for small to medium based on knowing, or being able to estimate accurately, the
size projects. 3. Complete COCOMO which gives a more size of the software to be developed. There is often great
detailed estimate for large projects. uncertainty in the software size. It may result in the
inaccuracy of estimation.
There are three modes of development.
1. Organic mode C. Function Point Analysis
* Relatively small simple software projects. The Function Point Analysis is a method of quantifying the
* Small team with good application experience work to a set size and complexity of a software system in terms of the
of less than rigid requirements. functions that the systems deliver to the user. A number of
Similar to previously developed projects. proprietary models for cost estimation have adapted to this
* Relatively small and require little innovation. type of approach, like as ESTIMACS and SPQR/20. This is a
2. Semi-Detached mode measurement which is based on the functionality of the
* Intermediate(in size and complexity) software projects in program. It was first introduced by Albrecht [1]. The total
which team with mixed experience level must meet a mix of number of FP depends on the counts of distinct in terms of
rigid and less than rigid requirements. format or processing logic types. Following two steps in
3. An embedded mode Software project that must be counting function points:
developed within a set of tight hardware software and
operational constraints. Counting to the user functions: The raw function counts
are arrived at by considering a linear combination of five
BASIC COCOMO basic software components. These components are external
Basic COCMO is an empirical estimation for estimating effort, inputs, external outputs, external inquiries, logic internal
cost, and schedule for software projects. It was derived from files, and external interfaces, each at one of three complexity
the large data set from 63 software projects ranging in size levels: simple, average or complex. The sum of these
from 200 to 100000 lines of code, and programming numbers, weighted according to the complexity level, is the
languages ranging from assembly to PL/I. This data was number of FC.
analyzed to discover a set of formula that was the best fit to
the observation these formula link the size of the system. In Adjusting for environmental processing complexity: The
COCOMO 81 effort is be calculated as PM= a*Size^6ΠEmi i=1 final function points are arrived at by multiplying function
to 15 Where a & b are the domain constant in the model. It count by an adjustment factor that is determined by
contains 15 effort multipliers. This estimation scheme considering 14 aspects of processing complexity. This
accounts the experience and data of the past projects which is adjustment factor allows the function count to be modified
extremely complex to understand and apply the same. by at most 35% or -35%.

INTERMEDIATE COCMO D. Linear Models Commonly these models have the

In 1997, an enhanced scheme for estimating the effort for simple structure and trace a clear equation as
software development activities, which is called as COCOMO below:
II. COCOMO II has some special features which distinguish EFFORT = a0 +Σ n ai xi i=0 Where, a1, a2 ,an are selected
for another one the uses of this method are very hidden and according to the information of project, only allowed values
its result usually accurate. In COCOMO II effort requirements for xi are -1, 0, +1.[16]
can be calculated as PM=a* size ^B*ΠEmi i=1 to 17 Where
E=B+0.01*ΣSFj j=1 to 5 COCOMO II is associated with 31 E. Seer-sem Models
factors LOC measures as the estimation variable, 17 cost This model has been proposed in 1980 by Galorath Inc[9].
drivers, 5 scale factor, 3 adaptation percentage of Most parameters in seer-sem are commercial and, business
modification, 3 adaptation cost drives and requirements projects usually use seer-sem as their main cost estimation
volatility. method. Size of the software is the most important feature in
seer-sem method and a parameter namely Se is defined as
The Detailed COCOMO Model effective size. Se is computed by determining the five
The detailed COCOMO model differs from the intermediate indicators: new size, existing size, ramp redesign and retest
model is only one major aspect. The detailed model uses as below:
different effort multipliers for each phase of a project. These Se=new size + existing Size(0.4 redesign + 0.25 reimp +
phase-dependent effort multipliers yield better estimates 0.35 retest) After computing the Se the estimated effort is
than the intermediate model[11]. calculated as below: EFFORT= td = D - 0.2 × (Se / Cte )0.4
Where D = relevant to the staffing aspects It is determined
B. Putnam Model based on the complexity degree in staffs structure. Cte is
The Putnam model is an empirical effort estimation model. computed according to the productivity and efficiency of the
Putnam used his productivity levels observations to derive project method. It is used widely in a commercial project. [7]

@ IJTSRD | Unique Paper ID – IJTSRD26511 | Volume – 3 | Issue – 5 | July - August 2019 Page 892
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
5. NON ALGORITHMIC METHODS considered as 1. Choosing analogy 2. Investigating
Non Algorithmic methods use some information about the similarities and differences 3. Examining of analogy quality.
previous projects which are similar to the underestimate 4. providing the estimation.
project is required and usually cost estimation process in
these methods is done according to the analysis of the B. Parkinson’s Law
previous datasets. Using Parkinson‟s Law “Work expands to fill the available
volume”[8], the cost is determined by the available resources
A. Expert Judgment Method Expert judgment techniques rather than based on an objective assessment., If the
involve consulting with cost estimation expert or a group of software has to be delivered in 20 months and 4 people are
the estimation experts to use their experience and available, the effort is estimated to be 80 PM. Although it
understanding of the proposed project to arrive at an sometimes gives a good estimation, this method is not
estimate of its cost. It is the most used methods for cost recommended as it may provide very unrealistic estimates.
estimation. Most companies used expert judgment method Parkinson‟s Law does not promote good software
for generating the cost of the product [4] [12]. This method engineering practice [2]. E. Price-to-win The cost is
using the following estimating steps: a. Project leader estimated to be the best price to win the project. The cost
presents each expert with a specification and an estimation estimation is based on the customer's budget instead of the
form. b. The experts fill out forms anonymously. c. Project software functionality. For example, if a reasonable
leader calls a group meeting in which the experts discuss estimation for a project costs 100 PM but the customer can
cost estimation issues with the project leader and each other. only effort 60 PM. It is common that the estimator is asked to
d. Project leader prepares and distributes a summary of the modify the estimation to fit 60 PM effort in order to win the
cost estimation on an iteration form. e. Again experts fill out project. This is again not a good practice since it is very likely
forms, anonymously. f. Steps d and step e are iterated for as to cause a bad delay of delivery or force the estimation team
many rounds as appropriate. to work overtime[10][2].

Estimation based on Expert Judgment is done by getting C. Top-Down Estimating Method

advice from experts who have extensive experiences in The top-down estimating method is known as Macro Model.
similar projects. This method is usually used when there is a Using this estimating method, overall cost estimation for the
limitation in finding data and gathering requirements. project is derived from the global properties of the software
Consultation is the basic issue in this method. One of the project, and then the project is partitioned into various low-
most common methods which work according to this level mechanism or components. The method of using this
technique in Delphi. Delphi arranges an especial meeting approach is the Putnam model. The top-down method is
among the project experts and tries to achieve the true more applicable to early cost estimation when only global
information about the project from there debates. Delphi properties are known. In the early phase of the software cost
includes some steps: estimation, top-down is very useful because there is no

The coordinator gives an estimation from each expert. detailed information available [4].

Each expert presents his own estimation

The coordinator gathers all forms and sums up them on D. Bottom-up Estimating Method
a form and ask experts to start another iteration. Using a bottom-up cost estimating method, the cost of each

Steps (ii-iii) are repeated until approval is gained. software component is estimated and then combines the
results to arrive at an estimated cost of the overall project.
A. Estimating by Analogy The bottom-up method aims at constructing the estimate of
Cost estimating by analogy means comparing the proposed a system from the knowledge accumulated about the small
project to previously completed similar project where the software components and their interactions. The method of
project development information is known. Actual data from using this approach is COCOMO's detailed model [4].
the completed projects are extrapolated to cost estimate the
proposed project. Analogy method can be used either at the 6. The Strength and Weakness of Algorithmic and Non-
system level or at the component level [12]. This method Algorithmic Software Cost Estimation Techniques
using the following estimating steps: a. Find out the According to the study of software cost estimation
necessary characteristics of the proposed project. b. Choose techniques, it is evident that no one method is necessarily
the most similar completed projects whose characteristics better or worse than the other, in fact, their advantage and
have been stored in the historical database. c. Find the disadvantage are often complementary to each other. The
estimate for the proposed project from the most similar algorithmic methods are based on mathematics and some
completed project by analogy. experimental equations. They are usually hard to learn and
they need much data about the current project state. But if
It means creating estimates for new projects by comparing enough data is available, these methods present reliable
the new projects to similar projects from the part. In this results. On the other hand for non-algorithmic methods, it is
method, several similar completed software projects are necessary to have enough information about the similar type
noticed and estimation of effort and cost are done according of previous projects, because these methods perform the
to their actual cost and effort. As the Algorithmic technique cost estimation by analysis of the historical data. They are
have a disadvantage of the need to calibrate the model. So easy to learn because they follow human behavior.
the alternative approach is ‘Analogy by Estimation’. According to the estimation experience, it is recommended
Estimation based on analogy is accomplished at the total that a combination of models and expert judgment
system levels and subsystem levels. By accessing the result estimation methods are useful to get reliable, accurate cost
of previous actual projects. We can estimate the cost and estimation for software development. We should use expert
effort of a similar project. The steps of this method are judgment method or analogy method for known projects and

@ IJTSRD | Unique Paper ID – IJTSRD26511 | Volume – 3 | Issue – 5 | July - August 2019 Page 893
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
projects parts if the similarities of them can be got since it is advantage of both the rigor of models and the speed of
fast and under this circumstance, reliable. For large, lesser- expert judgment method or analogy. Because the advantages
known projects, it is better to use algorithmic methods. In and disadvantages of each technique are complementary, a
this case, many researchers recommend the estimation combination will reduce the weakness of anyone technique,
models that do not require a source line of code as an input. augment their individual strengths and help to cross-check
If we approach cost estimation by parts, we may use expert one technique against another.
judgment for some known parts. In this way, we can take

Table1. The Strength of Algorithm and Non-Algorithmic Software Cost Estimation Techniques
Sr. No Method Type Strength
1 COCOMO Algorithmic Clear results, it’s very common
2 LOC Algorithmic Very easy in implementation to estimate the size of the software
3 Putnam model Algorithmic A Probabilistic model, it‟s used in a very large project
Seer-Sem
4 Algorithmic Used in very large projects
model
It‟s the best method of prediction using a linear regression
5 Linear model Algorithmic
technique
Works based on actual experience and especial expert is not
6 Analogy Non-Algorithmic
important
Expert
7 Non-Algorithmic Fast prediction, adapt for special projects
judgment
8 Parkinson Non-Algorithmic Correlates with some experience
9 Price to win Non-Algorithmic It‟s often gets the contract
Requires minimal project detail, usually faster and easier to
10 Top-down Non-Algorithmic
implement and system-level focus
More detailed basis, more stable and encourage individual
11 Bottom-down Non-Algorithmic
commitment

Table2. The Weakness of Algorithm and Non-Algorithmic Software Cost Estimation Techniques
Sr. No Method Type Weakness
1 COCOMO Algorithmic A lot of data is required, It is not suitable for any project
Prediction of the line is tough in the early stages, not good for a
3 LOC Algorithmic
very large project and language-dependent.
4 Putnam model Algorithmic For only use large projects
it‟s required 50 input parameters which are increases the
Seer-Sem model Algorithmic
5 complexity and uncertainty
Little difference between actual and predicted results and error
6 Linear model Algorithmic
is also needed to calculate.
Much information about past projects is required in some
7 Analogy Non-Algorithmic
situations there are no similar project
8 Expert judgment Non-Algorithmic Success depends on expert, usually is done incomplete
9 Parkinson Non-Algorithmic Reinforces poor practice
10 Price to win Non-Algorithmic Generally produces large overruns
11 Top-down Non-Algorithmic Less detailed basis and less stable
May overlook system-level costs, requires more effort, a lot of
12 Bottom -down Non-Algorithmic
time-consuming

7. ACKNOWLEDGMENTS methods are specific for some specific type of projects. It is

I would like to express my special thanks to all my teachers very difficult to decide which method is better than to all
who gave me their time and guidance, and all my friends other methods because every method or model has its own
who helped in the task of developing this paper. Finally, I significance or importance. Finding the most important
would like especially to thank my parents for their reason for the software project failure has been the object of
continuous support and encouragement throughout my many researchers in the last decade. According to the result
whole life. of this paper, the root causes for software project failure is
inaccurate estimation in stages of the project. To decrease
8. CONCLUSION the project failures software project managers are used to
In this paper, we have discussed a comparative study of select the best estimation method based on the different
different types of software cost estimation techniques and conditions and status of the project and also describe
also described the advantages and disadvantages of these comparing the estimation technique. There is no estimation
techniques. This paper presents some of the relevant method which could present the best estimates in all various
reasons that cause inaccurate estimation. To produce a situation and technique can be suitable in the special project.
meaningful and reliable cost estimate, we must improve our To improve the performance of the existing method and
understanding of software project attributes and their causal introducing the new methods for estimation based on
relationships. It has been seen that all cost estimation today’s software project requirements can be the future

@ IJTSRD | Unique Paper ID – IJTSRD26511 | Volume – 3 | Issue – 5 | July - August 2019 Page 894
 International Journal of Trend in Scientific Research and Development (IJTSRD) @ www.ijtsrd.com eISSN: 2456-6470
works in this area. The future work is to study the new Education Technology and Computer (ICETC), 2nd
software cost estimation technique that can help us to easily International Conference on, 2010.
understand the software cost estimation process.
[7] Li, J,J. Lin, et al. “ Development of the Decision Support
System for Software Project cost Estimation
References
Information Science and Engineering “. 2008. ISIS ’08.
[1] Leungh, Zhangf,‗”Software cost estimation” in
International Symposium on, 2008.
Handbook of software engineering and knowledge
engineering„ (World Scientific Pub. Co, River Edge, NJ, [8] G.N. Parkinson” Parkinson„s Law and Other Studies in
2001) Administration” Houghton-Mifflin, Boston, 1957.
[2] Sweta Kumari and Shashank Pushkar,” Performance [9] Galorath, D. D., & Evans, M. W. “ Software sizing,
Analysis of the Software Cost Estimation Methods”, estimation, and risk management: When performance
International Journal of Advanced Computer Science is measured performance improves”. Boca Raton, FL:
and Applications, Vol. 3, 2013. Auerbach,2006.
[3] Oscar Marbán, Antonio de Amescua, Juan J. Cuadrado, [10] Maged A. Yahya, Rodina Ahmad, Sai Peck Lee, “ Effects
Luis García “A cost model to estimate the effort of data of Software Process Maturity on COCOMO II’s Effort
mining projects”, Universidad Carlos III de Madri Estimation from CMMI perspective”, 978-1-4244-2379-
(UC3M), Volume33, Issue 1, pp.133-150, March, 2008 . 8/08 IEEE (c),2008
[4] B. W. Boehm”, Software Engineering Economics” [11] Leungh, Zhangf,‗” Software cost estimation” in
Prentice-Hall, Englewood Cliffs, NJ, USA, 1981. Handbook of software engineering and knowledge
engineering„ (World Scientific Pub. Co, River Edge, NJ,
[5] Attarzadeh, I. Siew Hock Ow, “Proposing a New
2001)
Software Cost Estimation Model Based on Artificial
Neural Networks”, IEEE International Conference on [12] Yahya, M. A., R. Ahmad et al. “ Effects of software
Computer Engineering and Technology (ICCET), process maturity on COCOMO II’s effort
Volume: 3, Page(s): V3-487 - V3-491 2010. estimation from CMMI perspective “. Research
Innovation and vision for the future, RIVF. IEEE
[6] Sikka, G., A. Kaur, et al. “ Estimating function points:
International Conference on, 2008.
Using machine learning and regression models”.

@ IJTSRD | Unique Paper ID – IJTSRD26511 | Volume – 3 | Issue – 5 | July - August 2019 Page 895