According to the research conducted by Hihn & Habib-agahi (1991), it is reviewed
that the survey conducted from the Jet Propulsion Laboratory staff, who evaluate
the software intensive projects costs in the technical divisions Jet Propulsion
Laboratory. Some of the respondents to the study defined all of the techniques
used by them in software costs estimation and, in case of the experiment, every
respondent projected the cost and size of a particular part of the software that
is defined in design document (Hihn & Habib-agahi, 1991). It was initiate
that technical staff majority estimating costs of software use informal correspondence
and high level of requirements segregating and that no formal process is here
for integrating uncertainty and risk. The technical staff of the Jet Propulsion
Laboratory is significantly good at effort estimation than the estimation of
size; though, in these case alterations are huge that there is the probability
of about 30% that anyone can evaluate and it is about 50% off (Hihn &
Habib-agahi, 1991).
The survey conducted in the paper had two main purposes,
one is to recognize the current practices and methods of costing within divisions
of Jet Propulsion Laboratory technical, and to get a clearer depiction of the accurateness
of the size and effort estimations. The most accurate approach that can be used
to accomplish these purposes would be to perceive persons when they make estimates
regarding cost and to gather data on estimations and real effort and size of
the project. The problem in using this approach is that it might take years to gather
adequate data for investigation, and it is also very expensive (Hihn &
Habib-agahi, 1991).
The last point is maybe the most significant, for programs of NASA, almost
every project has an important share of elements that are innovative. It means
that the estimation of anything that has an important number of unidentified objects
is a typical part of the environment estimation (Hihn & Habib-agahi, 1991).
A Review of Surveys on Software
Effort Estimation
According to the research conducted by (Moløkken
& Jørgensen, 2003 ) it is reviewed that the knowledge estimation
through surveys reviews on software estimation of effort. The major findings of
the article are that many of the projects about 60 % to 80% come across schedule
or effort overruns. The overruns conversely, appear to be less than the
overruns informed by some consultancy organizations. For instance, ‘Chaos
Report’ of Standish Group’s defines the cost overrun of 89 percent that is
higher as compared with the average overruns in other surveys that are 30% to 40% (Moløkken
& Jørgensen, 2003 ).
The methods of estimation in the most regular use are judgments
by experts. A potential purpose for the recurrent expert judgment use is that
there is no indication that recognized models of estimation lead to correct
estimates. There is also a deficiency of surveys with widespread analyses of causes
for schedule and effort overruns. It is hard
to maintain a balanced opinion on industry’s software estimation performance deprived
of impartial information from illustrative projects and companies (Moløkken
& Jørgensen, 2003 ).
This surveys in the paper also accessible in methodical
journals and discussions may be a basis for such impartial information. The
estimation results regarding surveys are also summarizes in the paper for software
estimation. It can also be said that there
has not been shown any organized estimation surveys review with the purpose of
summarizing software effort estimation performance knowledge (Moløkken & Jørgensen, 2003 ). An objective of this
survey is to discover how different projects are projected. For instance, in
case a project put on a model for estimation. The survey attempts to find the interpretation
of respondents regarding the estimations of efforts (Moløkken
& Jørgensen, 2003 ).
Software Project Effort and Cost
Estimation Techniques
According to the research conducted by (JyotiG.Borade
& Khalkar, 2013) it is reviewed that the highest objective
of effort and cost estimation of the software project is to methodically evaluate
the essential workload and the costs in the software system life cycle. The
cost estimation of the software is a very complex action that needs information
on many key features that affect the software projects outcomes, both independently
and in performance. The most serious issue that a lot of data is required that
is normally difficult to get in required measures. Therefore, effort and cost estimation
in the software has turned out to be a challenge for Information technology
industries (JyotiG.Borade & Khalkar,
2013).
In this article, numerous current methods for software project estimation cost
and effort are demonstrated, and their features are discussed. Likewise, it defines
metrics of software used for cost estimation in the software project. The paper
comprises remarks on the research trends description and estimation models performance
in cost estimation of software (JyotiG.Borade & Khalkar,
2013).
Likewise, presented information background on software project
software metrics and models used for cost and effort estimation. No model can estimate
the software cost with high relatively high accuracy. The estimation of the
effort and size of a software project is a multifaceted activity that needs key
attributes knowledge. At the preliminary project stage, there is high risk there
regarding the attributes of the project. It is also found that BBNs are particularly
beneficial when the evidence regarding the past and the current position is unclear,
imperfect, contradictory, and unclear (JyotiG.Borade & Khalkar,
2013).
The conventional estimation techniques emphasis on the actual growth effort; also,
this paper defined effort estimation of the test. About 40% of the total effort
in software development makeup by the testing activities (JyotiG.Borade
& Khalkar, 2013)
Software development cost
estimation approaches –A survey
According to the research conducted by (Boehma, Abtsa, & Chulani, 2000), it is reviewed that
numerous software cost estimation techniques and models classes such as expertise
based techniques, parametric models, dynamics-based models, learning-oriented
techniques, composite-Bayesian techniques, and regression-based models, for incorporating
regression-based and expertise-based models. Experience specifies that dynamics-based
and neural-net techniques are less developed as compared with other techniques classes,
but all techniques classes are tested by the rapid software technology change.
The main assumption is that not one technique is best for all of the states and
that a cautious contrast of the several approaches results that are most likely
to produce more accurate estimates (Boehma, Abtsa, & Chulani,
2000).
Like other fields, the cost models of the software
engineering field had its drawbacks. The rapidly changing software development nature
has made it hard to grow parametric models that produce high software
development accuracy in all fields. The costs of the software development endure
to growth and practitioners frequently express their concerns about incapability
to correctly forecast the costs (Boehma, Abtsa, & Chulani,
2000).
This article also presented the software estimation techniques overview, offering
an indication of numerous standard estimation models now obtainable. Experience
shows that dynamics-based and neural-net techniques are less advanced than
other techniques classes, but that all the overall techniques classes are
tested by the rapid software technology change. The main assumption is that not
one technique is best for all of the states and that a cautious contrast of the
several approaches results that also estimates one might vary considerably than
the other estimates about the class techniques.
Software Project Effort: Different
Methods of Estimation
According to the research conducted by (Hamdan &
Madi, 2011 )
it is reviewed that the research is more focus on the measure and, significantly,
the software development prediction projects as far as the estimation of the cost
is the concern. The main objective of the paper is to discover the recording usefulness,
in the data of software project, the leadership and cultural features of the team
development. As an outcome of this research, an innovative model for recognizing
and investigating was established. The paper examines the relationship of these
features with other characteristics of a project using information from a study
of forty-one software projects gathered from an information technology
organizations in the United Arab Emirates (Hamdan & Madi, 2011 ).
The background and culture of the leader affect his
decision-making and the way the organization is controlled. It helps with the
appropriate working of an organization and the worker’s aptitude to move from
one style of leadership to another. The management in the organization can have
an impact on the way projects are operational and the estimate of cost for
projects with analogous needs, and with dissimilar leaders in the organization.
For instance, consider the software team quality (for example compatibility and
capability of analysts and programmers). Some of the researchers have recurrently
confirmed that the leadership support missing in a project is normally a reason
for the ultimate failure of the project (Hamdan & Madi, 2011 ).
The background and culture of the leader in the team is a
significant factor in defining the quality or cost of the resultant products in
software. Through statistical and surveys analyses, the research recognized the
important leadership and culture influences and expanded a background to
support the application for an environment of effective work. The results of
the study suggest that good estimations are gained when leadership and cultural
qualities are comprised in the estimation model. Specifically, the actual
effort estimation improved significantly for both core and support systems,
when attributes related to cultural and leadership were added (Hamdan &
Madi, 2011 ).
Prediction of software
development effort estimation using neural networks
According to the research conducted by (Kumar & K, 2016), it is reviewed that
the software failures are mainly because of the faulty practices of project
management, that include estimation of effort. Constant software development
technology changing outlines make the estimation of the effort more stimulating.
Numerous approaches are available to evaluate the effort between which method
based on soft computing plays a protuberant role. The deals of software cost
estimation with a lot of hesitation between all methods of soft computing
neural network is effective in uncertainty management. This paper also
recommends a BPNN to use and improves the estimation of effort for data set of Cocomo.
In this research MMRE & MRE are used as the criteria of evaluation. The estimation
of effort is a procedure of forecast feasible development time and cost to make
a software product. Precise estimation of effort is significant as over
estimation result in the business loss and under estimation result in the software
with low quality that rapidly leads to failure of software (Kumar & K, 2016).
There are different neural network methods have been
used to analyze estimated effort. All of the technique emphases on providing
best effort estimation for the software. In this paper, it is found that a neural
network is an effective upcoming estimating effort development. It was recommended
for computationally and huge complex projects; it is better to use a neural
network method. There is a need to inspect methods accuracy that mainly needed
in effort estimation of the software (Kumar & K, 2016). The neural network performance
depends on its design and their settings of a parameter. There are numerous considerations
control the design of the neural network as well as total layers, total nodes of
each layer and in each node the transfer function (Kumar & K, 2016).
A Systematic Review of Software
Development Cost Estimation Studies
According to the research conducted by (Jørgensen
& Shepperd, 2007), it is reviewed that the goals to offer
a source for the software estimation improvement research via a systematic
review. The review recognizes cost estimation software papers in about
seventy-six journals and categorizes the identifications according to the
estimation approach, research topic, study context, roach data set and research
app. A library of cost estimation documents is offered to ease the documentation
of appropriate research results estimation (Jørgensen & Shepperd, 2007). The results of the review
mutual with other information offers support for suggestions for future research
of software cost estimation, comprising the growth in the breadth of search for
appropriate studies, searching for papers within a carefully selected journals when
extensiveness is important and it is also important to conduct estimation
methods usually used by industry of software (Jørgensen & Shepperd, 2007).
There is a deficiency of consistent use of terms relating
to cost estimation software. It is believed that this lack makes it is easy for
them to miss papers when depending on digital libraries automatic searches. The
search term “cost estimation soft-ware” for instance, did not recognize more
than 60 % papers on cost estimation of software recognized by manual search (Jørgensen
& Shepperd, 2007). It is also likely to increase the responsiveness
of how data sets properties have an influence on the outcomes while assessing
estimation approaches. Presently, a typical assessment of the estimation technique
depends on the randomly selected set of data, where representativeness and
properties are not discussed and examined (Jørgensen & Shepperd, 2007).
A review of studies on expert
estimation of software development effort
According to the research conducted by (Jørgensen M.
, 2004)
it is reviewed that the main objective and influence is to support the expert
estimation research, for example, to
ease the search of researcher’s for related studies of expert estimation. In accumulation,
it also offers software experts with valuable guidelines estimation, depends on
the knowledge based on research of expert processes of estimation. The evaluation
results recommend that estimation of expert is most often practical software
projects estimation strategy, and there is no considerable indication in favor
of estimation models use, and there are circumstances where it is expected from
the experts to estimates more accurate as compared with the formal models of estimation
(Jørgensen M. , 2004). Some of the best
expert estimation ‘‘guidelines are also found in this paper, that is to
evaluate the accuracy of estimation, but avoid high pressure of evaluation;
avoid goals of conflicting estimation; request the estimators to criticize and justify
their estimations and avoid unreliable and inappropriate information estimation (Jørgensen M.
, 2004);
For better estimation, it is also important to
find experts of estimation with related domain and good records of estimation. Estimate
bottom-up and top-down, individually and use checklists of estimation. It is also good to collect the estimates from
different estimation strategies and experts; the uncertainty of the estimation
also required to be done. (Jørgensen M. , 2004). There are circumstances
where estimates of the expert are possible to be more correct, for example the situations
where experts have significant knowledge of domain not comprised in the situations
or models when simple strategies of estimation offer correct estimates. The principles
of estimation to also depends on the results from other fields than software
development, or signify just one category of software experts and projects (Jørgensen M.
, 2004).
References
Boehma, B., Abtsa, C., & Chulani, S.
(2000). Software development cost estimation approaches –A survey . Annals
of Software engineering, 177-205.
Hamdan, K., & Madi, M. (2011 ).
Software Project Effort: Different Methods of Estimation. IEEE , 15- 18.
Hihn, J., & Habib-agahi, H. (1991).
Cost Estimation of Software Intensive Projects:A Survey of Current Practices. IEEE,
275- 287.
Jørgensen, M. (2004). A review of studies
on expert estimation of software development effort. The Journal of System
and Software, 70, 37–60.
Jørgensen, M., & Shepperd, M. (2007).
A Systematic Review of Software DevelopmentCost Estimation Studies. IEEE
TRANSACTIONS ON SOFtWARE ENGINEERING, 33(1), 33-53.
JyotiG.Borade, & Khalkar, V. R.
(2013). Software Project Effort and Cost Estimation Techniques. International
Journal of Advanced Research in Computer Science and Software Engineering,
730-739.
Kumar, T. M., & K, Y. T. (2016).
Prediction of software development effortestimationusing neural networks. International
Journal of Scientific Development and Research, 1(5), 430-434.
Moløkken, K., & Jørgensen, M. (2003 ).
A Review of Surveys on Software Effort Estimation . IEEE .
