Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)

IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

A Review of Effort Estimation in Agile Software

Development using Machine Learning Techniques
Sandeep Kumar M ohit Arora*
.
Dept. of Computer Science and Dept. of Computer Science and Sakshi
Engineering Engineering Dept. of Computer Science and
Lovely Professional University Lovely Professional University Engineering
Phagwara, Punjab, India Phagwara, Punjab, India Lovely Professional University
[email protected] [email protected] Phagwara, Punjab, India
[email protected]
Shivali Chopra
2022 4th International Conference on Inventive Research in Computing Applications (ICIRCA) | 978-1-6654-9707-7/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICIRCA54612.2022.9985542

Dept. of Computer Science and

Engineering
Lovely Professional University
Phagwara, Punjab, India
[email protected]
*corresponding author are assigned for each work item and based on the
requirement of effort to comp lete that work the value of the
Abstract— Nowadays organizations are heading towards points are decided. But it has been observed that average
the agile software development instead of traditional error [2] of 20% to 30% is there in effort estimation using
development models due to the flexibility and acceptance of story points so to reduce that error different machine learning
change in the agile software development approach. Estimation and neural network techniques are used along with story
of time, effort, and size of a project is a challenging task in points.
agile software development due to the changing requirement
during development. S o to overcome the challenges in effort II. LIT ERAT URE REVIEW
estimation different Machine Learning (ML) techniques such
as Random Forest, Decision tree, S GB, and Neural Network Shashank Mouli Satapathy et al.[2] Used different
models are used along with story points. In this paper, a machine learn ing techniques s uch as Random Forest, SGB,
literature review of all these ML techniques is carried out for and Decision tree for effort estimation of an agile project.
effort estimation of agile projects. The evaluation of the various model’s performance is
accessed by MAE (Mean Absolute Error), PRED (The
Keywords— Machine Learning Techniques, Agile S oftware Prediction Accuracy), and MMER (Mean of Magnitude of
Development, S tory Points, Neural Networks, Effort Error Relative). SGB provides quite high accuracy out of
Estimation. these techniques.

I. INT RODUCT ION Aditi Panda et al.[3] Discussed different neural network
techniques such as Probabilistic Neural Network (PNN),
Organizat ions are heading towards agile software General Regression Neural Network (GRNN), Group
development instead of traditional software development Method of Data Handling (GMDH), Cascade Correlation
models. Traditional software development model such as Neural Network, and Polynomial Neural Network. In this
waterfall, iterative etc. are providing very less success rate in study proposed steps predict the effort estimation of agile
software development due to the rigidity in models for user software by applying all the neural network techniques. After
requirement and by that customers are not fully satisfied with comparing the outcome it is found that cascade correlation
those projects as customer involvement is not there in the network provides better results than rest of the networks, and
development process. In agile software development the GRNN is providing the worst results.
success rate of projects has been improved a lot due to the
adoption to change and customer involvement during the Peeka Abrahamsson et al.[4] Discussed that development
development process. effort is predicated using user story. Proposed methodology
is applied on two agile projects of different size and
Agile Software Develop ment is introduced to overcome structure. It observed that if user storage is written in
the challenges of frequently changing requirement during the structural way than effort estimation provides good result.
software development. Due to the iterative and incremental
approach of the agile software development [1] an Shashank Mouli Satapathy et al.[5] explained a
improvement is made in the project development by the methodology to achieve better accuracy in prediction where
customer evaluation and feedback. Agile software Support vector regression technique based on different kernel
development help organizations to produce software with methods (SVR Linear Kernel, SVR RBF Kernel, SVR
flexible requirements as agile software development allows Polynomial Kernel, and SVR Sig moid Kernel) are used.
for flexibility during project development. Then results are compared and performance is evaluated. Out
of these, RBF Kernel technique provides the best res ults.
Since estimation of effort and resources is major
objective at the start of project development and to achieve Poonacha K.M. et al.[6] Explained as software
this goal traditionally different techniques [2] such as development is rapidly moving towards agile so it is
analogy based estimation and expert-opinion are used but the important to measure the level of the agility for an
results of these techniques can be biased or inaccurate. For organization. Client will also be interested to get the
successful completion of ag ile project the estimation of time, information about the agility level. Survey is conducted to
effort, size and cost should be accurate. Story point is used to access the performance of the software house against each
get the effort estimation values in agile projects where points parameter. A HP (Analytic Hierarchy Process) is used to

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 416

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)
IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

analyze the data received through survey. Then this received teams. So, in this paper a methodology of agile story point is
data is used to create the training and testing datasets for proposed that ensures the relativity in software sizing during
ANFIS (Adaptive Network Based Fuzzy Interference System the use of agile story points and can be adopted by CMMI
Engine). ANFIS is used to measure the agility of data. Using organizations. Proposed methodology is imp lemented on
AHP, Quantification process can be simplified by the CMMI level-3 Company and it is observed that the error
decomposition of problem into set of pairs that can be solved percentage in effort estimation is found as 5.9% but earlier it
easily. Basic ANFIS framework makes it very complicated was found as 28% error when other techniques were used.
so we are using AHP-ANFIS framework, by this complexity
Rashmi Popli et al.[13] Discussed about the need of
of 17 parameters are converted into four group parameters.
Then agility is measured with the help of ANFIS. priorit izing the user stories in agile development as most of
the previous prioritization methods doesn’t consider the
Saru Dhir et al.[7] Proposed a model for accurate importance of client’s user stories. In proposed methodology,
estimation of agile projects. In the proposed model, two researcher also proposed some importance related factors and
techniques i.e. Feed-forward and Feedback approach are effort related factors along with algorithm to prioritize user
used along with the agile techniques in the process cycle to stories.
maintain the reliability of the end product. Then the proposed
model is imp lemented in different phases. Also an Ziauddin et al.[14] Exp lained the different traditional
techniques of estimation and also discussed about agile
algorithmic estimation approach is explained and a new EPIS
software development. The proposed methodology is created
approach is discussed. EPIS depends on different parameters
such as experience of team members, estimation parameters, with help of story size, and agile velocity. Further these
values are used to calculate the development cost, effort, and
and feed forward of similar projects. Estimated velocity can
be improved based on the feed forward and past experience time. An experimental analysis has been carried out with the
help of the data of 21 developed projects and it is observed
used by the senior members. Further these techniques (i.e.
Algorithmic and proposed model) are compared with the that time estimation results are predicted to be 57.14%
lowers and cost estimation results are 61.9% lo wer than
traditional existing approaches.
previously calculated results.
Hind Zahraoui et al.[8] Discussed the effort estimat ion of
a scrum project using story points and also highlighted the Saurabh Bilgaiyan et al.[15] Proposed effort estimation in
agile software with the help of t wo neural network
inaccuracy in estimation using this approach. Certain factors
such as priority factor, story size factor and complexity techniques i.e. Feed-forward back propagation network and
Elman network. These two models are applied on 21 project
factor are responsible for the uncertainty in the estimat ion of
a project using story point. To avoid these uncertainties story dataset and their performance are measured using various
parameters such as MSE,MMRE,PRED. Results of various
point adjustment factor (SPF) is calculated and using SPF the
Neural Network models are co mpared and it is observed that
adjusted story point is calculated. Further to get accurate
estimation, adjusted story point is used instead of story point. effort estimation of Feed-forward back propagation network
provides the best effort estimation results.
Evita Coelho et al.[9] Has given an overview of the
different estimation techniques used traditionally such as Atef Tayh Raslan et al.[16] Proposed a framework
consisting of COCOMO 2.0, story points and fuzzy logic to
SLOC, FPA, and use case points method. In this paper
researcher proposed a technique based on story point to improve the effort estimation accuracy. In proposed
framework estimat ion is done in two phases as preliminary
calculate the effort estimation using Different factors such as
Estimation of user stories, customer expectations, and and constructive iteration phase using dataset from six
NASA centers that includes 10 pro jects. After comparing the
estimation of delivery date. Effort estimation based on story
results it is observed that proposed framework improves the
points uses the concept of velocity to measure the progress
rate of the team. prediction value up to 8%.
Abhishek saini et al.[17] Introduced a fuzzy model to
Jitendra Choudhari et al.[10] Introduced a software
maintenance effort estimation model (SM EEM) to estimate calculate the effort estimation of agile software development.
Raw data of the actual figures are used as input in fuzzy
software maintenance. Researcher has also discussed about
the different factors that affects the software maintenance model and output is received as estimated effort. Three types
of input variables are used i.e. user story, complexity, and
estimation. The proposed model has been implemented and
team expert ise. Based on these input parameters fuzzy ru les
evaluated as ASP computation. This model is suitable for
only Agile and XP based maintenance estimation. are derived. Further to measure the estimation, data of 6
projects are taken and implemented. Then results are
Atef Tayh Raslan et al.[11] Proposed a framework for the categorized as low, medium, and high efforts.
effort estimation using some fuzzy logic and story points in a
Hala Hamad Osman et al.[18] Explained the different
agile software development. In this paper a brief introduction
of fuzzy logic has been provided. In the proposed framework estimation methods of agile software such as story point,
planning poker, and expert opinion. It is observed that none
fuzzy input parameters are received as story point, friction
of these single techniques is best for all situations. So there is
factors, implementation level factor, and dynamic forces.
Then this input will be processed in further steps to get t he a need for new technique such as combination of FP and
story points, combination of planning poker with some other
final effort estimation. The proposed framework is designed
using MATLAB. technique, Factor-Based estimation in agile software, Soft-
computing technique in agile and also Neuro-fuzzy
Alaa El-deen Hamouda et al.[12] Explains how it is techniques. These new proposed techniques provide better
difficult for a CMMI organization to accept the agile results than single estimation technique used.
software estimat ion due to the inconsistency in value of story
points within the organization observed by different project

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 417

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)
IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

Marek Milosz et al.[19] Discussed that story point Amrita Sharma et al. [28] found the correlation between
method is considered as main method for calculation of the dependent and independent variables. Three multip le linear
development estimation. Team velocity is estimated in this regression models were developed and these models were
study and case study was performed to evaluate the results. evaluated using MMRE values. Zia’s dataset is used for
After the research of one semester, it is observed that in evaluation and performance of these linear models are
execution of single iteration, value of velocity is ranging compared with DT, SGB, RF. Using MMRE values it has
fro m 60% to 160%. Researcher also mentioned the been observed that model1 provides better accuracy out of
advantages of using story point method over other methods. these techniques.
Ch. Prasada Rao et al.[20] Imp lemented machine Mohit Arora et al.[29] Carried out a start of art regression
learning techniques such as adaptive Neuro-Fuzzy modeling, techniques review. Six different regression techniques i.e.
Radial Basis Function Network, and Generalized Regression LR, DT, RF, xg Boost, CatBoost and AdaBoost are
Neural Net work (GRNN) along with story points for effort implemented on Zia’s dataset. Performance of these
estimation. Results of thes e ML Techniques are calculated regressor models is co mpared and it has been observed that
using MATLAB with the help of 21 project’s data. It is CatBoost outperforms other regressor models.
observed that GRNN provides the better results out of these
techniques due to the high PRED value and low MMER, M Vyas et al.[30] Proposed the regression based
techniques for effort estimation of agile projects. Three types
MMRE values.
of regression models i.e Linear, Ridge and Logistic
Thanh Tung Khuat et al.[21] Proposed an algorithm for regression model are used on zia’s dataset. After comparing
the accuracy of agile effort estimation using artificial neural the performance of these models it has been observed that
networks along with fireworks algorith m. After applying Ridge regression model provides better results than other
proposed FWA-LM algorith m on 21 pro ject datasets, results regression models.
are being compared with other neural network and regression
models. After co mparing performances of all models, it has Amrita Sharma et al. [31] carried out a comparative study
of effort estimation of agile development and traditional
been observed that FWA-LM outperforms other neural
development using neural network (NN) and genetic
network models.
algorithm (GA). Error values( using different evaluation
Anupama Kaushik et al.[22] Discussed the DBN-A LO parameters such as MMRE, MSE, and R2 ) of NN and GA are
technique for the effort estimation of agile and non -agile calculated for project based on LOC and project based on SP
projects. Three agile datasets i.e. 21 project dataset, CD-1, separately. Observed values of MMRE, MSE and R^2 are
CD-2 are used for agile estimation. After applying DBN - compared.
ALO on these dataset it has been observed that this technique
Ravi Kiran Mallidi et al.[32] Discussed the agile scrum
provides best results for both agile and non-agile projects.
models for effort estimat ion. Different scrum models such as
Mohit Arora et al.[23] Discussed the need for the effort Planning Poker, T -Shirt size, Dot Voting, Bucket System,
estimation in agile pro jects and reviewed the existing ordering methods etc. are described along with the
techniques of agile effort estimation. Different machine challenges faced in each effort estimation technique of agile
learning techniques used for effort estimation in agile is projects.
mentioned along with MMRE values of those techniques. It
has been observed that ML Techniques provides better Laura- Diana Radu et al.[33] Proposed an effort
estimation model for agile software development using
results than NON – ML and traditional techniques.
Bayesian networks. Based on the literature review it has been
Emanual Dantas et al.[24] Updated the systematic identified that two major factors that affect effort estimation
literature review of 2014. Forward snowballing approach is are teamwork quality and user stories characteristics.
applied using previous SLR and related papers and it has
been observed that there is a gap of suitable cost drivers in Helcyon D. P. Carvalho et al.[34] Proposed eight
ensemble regression models for effort estimat ion of agile
previous SLR. So in this updated review, themat ic analysis is
applied and a representative set of 10 cost drivers are software development. These proposed models i.e. B-RR, B-
LR, B-LA, B-RI, ST-RR, ST-LR, ST-RI, ST-LA are based
identified for effort estimation.
on bagging and stacking methods.
P. Suresh Kumar et al.[25] Discussed the overview of
Yasir mah mood et al.[35] investigated the different
various ANN techniques for effort estimation. It has been
machine learn ing ensemble techniques and there
observed that ANN techniques provide better accuracy as
compare to the traditional models in effort estimation. performance is obtained. For performance observation two
parameters i.e. MM RE and PRED (25) are used. It has been
Marta Fernandez Diego et al.[26] Updated the literature observed that an ensemble effort estimation technique
review of usman et al.’s 2014 systematic literature review provides better results than solo techniques.
with help of 73 new papers. An increasing accuracy in the
effort estimation is being observed with the increasing Manju Vyas et al.[36] Carried out a review of effort
estimation techniques of agile software development.
number of papers.
Machine learning and Non- Algorithmic techniques are
Mohit Arora et al.[27] Carried out the literature review of compared and there advantages and disadvantages are also
various effort estimation techniques and proposed a discussed.
methodology to provide better estimation accuracy in agile
projects using soft computing techniques. Proposed Mohit Arora et al.[37] Proposed an effort estimation
technique using regression testing for agile software
methodology is divided into three categories i.e. Data
preparation, Data set Partitioning and Model selection & development including story point approach. Proposed
testing part.

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 418

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)
IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

technique is imp lemented using four project data and it has factors are categorized as dependent and independent
been observed that obtained results are near to actual values. variables. Independent variables further consists of
organizational factors and environmental factors whereas
Prerna shakya et al.[38] discussed the importance of dependent variables consists technical factors, project
factors, team factor, and customer factor.
critical success factors in agile software development .
Researcher proposed a framework that focuses on the
critical success factors of ag ile in Nepal industries. These

Table 1
Literature Review Summary

Author Name Algorithm Used Dataset Accuracy Evaluation Criteria Year

Shashank Mouli Random Forest (RF) Zia RF: MMER=0.2516, MMER, 2017
Satapathy et al. PRED(x)=95.2381 MdMER,
Decision tree (DT) DT: MMER=0.3820, PRED(x)
PRED(x)=90.4762
Stochastic gradient SGB: MMER=0.1632,
boosting(SGB) PRED(x)=95.2381

Aditi Panda et General Regression Zia GRNN: MSE, 2015

al. Neural Network MMRE=0.3581, MMRE,
PRED(x)=85.91 R^2(coefficient),
Probabilistic Neural PNN: PRED(x),
Network MMRE=1.5776
PRED(x)=87.65
Group Method of Data GMDH:
Handling MMRE=0.1563,
PRED(x)=89.6689
Cascade Correlation CCNN:
Neural Network MMRE=0.1486,
PRED(x)=94.7649

Shashank Mouli SVR Linear Kernel Zia SVR Linear: MMRE, 2014
Satapathy et al. MMRE=0.1492, PRED(x)
PRED(x)=90.8112
SVR Polynomial SVR Polynomial:
Kernel MMRE=0.4350,
PRED(x)=68.7382
SVR RBF Kernel SVR RBF:
MMRE=0.0747
PRED(x)=95.9052
SVR Sigmoid kernel SVR Sigmoid:
MMRE=0.1929,
PRED(x)=89.7646

Alaa El-deen ML model using Level 3 Size estimation error decreased MER 2014
Hamouda et al. Neural network CMMI from 28% to 5.9%,
organization Effort estimation error
decreased from 34% to 12%.

Ziauddin et al. Effort estimation Zia MMRE(Time)=0.0719 MMRE, 2012

model using story MMRE(Cost)=5.76% PRED
points PRED Time(7.19)=57.14%
PRED Cost(5.76)=61.90%

Saurabh Feed-forward Back Zia FFBP Network: MMRE, 2019

Bilgaiyan et al. Propagation Network MMRE=0.1349 MSE,
MSE=0.052 PRED(x)
PRED(25)=95.2301

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 419

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)
IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

Elman NN Elman NN:

MMRE=0.1480
MSE=0.056
PRED(25)=94.8659

ANFIS ANFIS: MMRE, 2018

MMRE(cost)=3.9079 MMER,
MMRE(time)=8.4277 PRED
PRED(cost)=57.1429
PRED(time)=76.1905
GRNN GRNN:
MMRE(cost)=4.8385
MMRE(time)=2.7864
Zia PRED(cost)=76.1905
Ch. Prasada Rao PRED(time)=76.1905
et al.
RBFN RBFN:
MMRE(cost)=9.9604
MMRE(time)=8.0909
PRED(cost)=76.1905
PRED(time)=76.1905

Thanh Tung ABC-PSO Zia MMRE=0.0569 MMRE 2017

Khuat et al MdMRE=3.33% MAR
PRED(8)=66.67 PRED
R^2=0.9734 R^2
MAR=3.12
Anupama DBN-ALO Zia, MMRE=0.0225 PRED(25), 2019
Kaushik CD1, MdMRE=0.0222 MMRE,
CD2 PRED(25)=98.4321 MdMRE
Thanh Tung FWA-LM Zia MMRE=0.0293 MMRE, 2016
Khuat R^2=0.9946 R^2,
PRED(7.19)=100 PRED,
MSE=3.7983 MSE
Amrita Sharma Linear regression Zia MMRE= 0.099 MMRE 2020
model 1
M Vyas Ridge regression Zia Ridge Regression: MMRE, 2021
MMRE=0.13 PRED(25)
PRED(25)=85.71
Linear regression Linear Regression:
MMRE=0.15
PRED(25)=71.42
Logistic regression Logistic Regression:
MMRE=0.19
PRED(25)=71.42
Mohit arora et Linear regression Zia LR: PRED, 2021
al. PRED=0.9345 RMSE,
RMSE=5.8523 R^2
R^2=0.93
Decision Tree DT:
PRED=0.9215
RMSE=6.4070
R^2=0.92
Random Forest RF:
PRED=0.9179
RMSE=6.5505
R^2=0.9179

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 420

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)
IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

XGBoost XGBoost:
PRED=0.9427
RMSE=5.4745
R^2=0.9427
CATBoost CATBoost:
PRED=0.9852
RMSE=2.7784
R^2=0.9852

ADABoost ADABoost:
PRED=0.9511
RMSE=5.0555
R^2=0.9511

The values of mean magnitude of relative error (MM RE) better predictions. DBN-A LO has the lower values of
have been diagrammatically represented in Fig. 1 for the MMRE as co mpared to other algorith ms in the present
different algorith ms applied in the effort estimation in agile literature.
software development. Lower value of MM RE signifies the

MMRE
Logistic Regression
Linear Regression
Ridge Regression
Linear Regression Model 1
FWA-LM
DBN-ALO
ABC-PSO
Elman NN
FFBP Network
SVR RBF Kernel
SVR Sigmoid Kernel MMRE
SVR Polynomial Kernel
SVR Linear Kernel
CCNN
GMDH
PNN
GRNN
SGB
DT
RF
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
.
Fig. 1. MMRE comparison chart

III. CONCLUSION provides better effort estimation accuracy as compare to

other techniques.
A literature review of various estimation techniques is
underway and it has been recognized that measuring
efforts in the development of agile software is a challenge.
REFERENCES
Therefor to find the exact estimation values in agile
[1] H. Alahyari, R. B. Svensson, and T. Gorschek, “PT US CR,” J.
software development different machine learning Syst. Softw., 2016, doi: 10.1016/j.jss.2016.12.007.
techniques are used along with the story points. Further [2] S. M. Satapathy and S. K. Rath, “Empirical assessment of
machine learning models for agile software development effort
these ML techniques are implemented with different
estimation using story points,” Innov. Syst. Softw. Eng., vol.
datasets to get their effort estimation values and results are 13, no. 2–3, pp. 191–200, 2017, doi: 10.1007/s11334-017-
compared. After comparing MMRE values of different 0288-z.
[3] A. Panda, S. M. Satapathy, and S. K. Rath, “Empirical
techniques in Fig. 1 it has been observed that DBN-A LO Validation of Neural Network Models for Agile Software
Effort Estimation based on Story Points,” Procedia Comput.

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 421

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.
 Proceedings of the International Conference on Inventive Research in Computing Applications (ICIRCA 2022)
IEEE Xplore Part Number: CFP22N67-ART; ISBN: 978-1-6654-9707-7

Sci., vol. 57, pp. 772–781, 2015, doi: [23] M. Arora, S. Verma, and S. Chopra, “of Machine Learning
10.1016/j.procs.2015.07.474. Estimation Approaches in Scrum Projects.”
[4] P. Abrahamsson, I. Fronza, R. Moser, J. Vlasenko, and W. [24] E. Dantas, M. Perkusich, E. Dilorenzo, D. F. S. Santos, H.
Pedrycz, “Predicting development effort from user stories,” Almeida, and A. Perkusich, “Effort Estimation in Agile
Int. Symp. Empir. Softw. Eng. Meas., pp. 400–403, 2011, doi: Software Development: An Updated Review,” Int. J. Softw.
10.1109/esem.2011.58. Eng. Knowl. Eng., vol. 28, no. 11–12, pp. 1811–1831, 2018,
[5] S. M. Satapathy, A. Panda, and S. K. Rath, “Story point doi: 10.1142/S0218194018400302.
approach based agile software effort estimation using various [25] P. S. Kumar, H. S. Behera, K. Anisha Kumari, J. Nayak, and
SVR kernel methods,” Proc. Int. Conf. Softw. Eng. Knowl. B. Naik, “Advancement from neural networks to deep learning
Eng. SEKE, vol. 2014-Janua, no. January, pp. 304–307, 2014. in software effort estimation: Perspective of two decades,”
[6] K. M. Poonacha and S. Bhattacharya, “Towards a framework Comput. Sci. Rev., vol. 38, p. 100288, 2020, doi:
for assessing agility,” Proc. Annu. Hawaii Int. Conf. Syst. Sci., 10.1016/j.cosrev.2020.100288.
pp. 5329–5338, 2012, doi: 10.1109/HICSS.2012.599. [26] M. Fernández-Diego, E. R. Méndez, F. González-Ladrón-De-
[7] S. Dhir, D. Kumar, and V. B. Singh, “An estimation technique Guevara, S. Abrahão, and E. Insfran, “An update on effort
in agile archetype using story points and function point estimation in agile software development: A systematic
analysis,” Int. J. Process Manag. Benchmarking, vol. 7, no. 4, literature review,” IEEE Access, vol. 8, pp. 166768–166800,
pp. 518–539, 2017, doi: 10.1504/IJPMB.2017.086933. 2020, doi: 10.1109/ACCESS.2020.3021664.
[8] H. Zahraoui and M. A. Janati Idrissi, “Adjusting story points [27] M. Arora, S. Verma, and Kavita, “An efficient effort and cost
calculation in scrum effort & time estimation,” 2015 10th Int. estimation framework for Scrum Based Projects,” Int. J. Eng.
Conf. Intell. Syst. Theor. Appl. SITA 2015, 2015, doi: Technol., vol. 7, no. 4.12 Special Issue 12, pp. 52–57, 2018,
10.1109/SITA.2015.7358400. doi: 10.14419/ijet.v7i4.12.20992.
[9] R. Popli and N. Chauhan, “Cost and effort estimation in agile [28] A. Sharma and N. Chaudhary, “Linear Regression Model for
software development,” ICROIT 2014 - Proc. 2014 Int. Conf. Agile Software Development Effort Estimation,” 2020 5th
Reliab. Optim. Inf. Technol., vol. 3, no. 7, pp. 57–61, 2014, IEEE Int. Conf. Recent Adv. Innov. Eng. ICRAIE 2020 -
doi: 10.1109/ICROIT.2014.6798284. Proceeding, vol. 2020, pp. 4–7, 2020, doi:
[10] J. Choudhari and U. Suman, “Story Points Based Effort 10.1109/ICRAIE51050.2020.9358309.
Estimation Model for Software Maintenance,” Procedia [29] M. Arora, A. Sharma, S. Katoch, M. Malviya, and S. Chopra,
Technol., vol. 4, pp. 761–765, 2012, doi: “A State of the Art Regressor Model’s comparison for Effort
10.1016/j.protcy.2012.05.124. Estimation of Agile software,” Proc. 2021 2nd Int. Conf.
[11] A. Tayh, R. D. Nagy, and H. A. Hefny, “Towards a Fuzzy Intell. Eng. Manag. ICIEM 2021, pp. 211–215, 2021, doi:
based Framework for Effort Estimation in Agile Software 10.1109/ICIEM51511.2021.9445345.
Development,” IJCSIS) Int. J. Comput. Sci. Inf. Secur., vol. 13, [30] M. Vyas and N. Hemrajani, “Predicting effort of agile software
no. 1, pp. 37–45, 2015, [Online]. Available: projects using linear regression, ridge regression and logistic
http://sites.google.com/site/ijcsis/. regression,” Int. J. Tech. Phys. Probl. Eng., vol. 13, no. 2, pp.
[12] A. E. D. Hamouda, “Using agile story points as an estimation 14–19, 2021.
technique in CMMI organizations,” Proc. - 2014 Agil. Conf. [31] A. Sharma and N. Chaudhary, “Analysis of Software Effort
Agil. 2014, pp. 16–23, 2014, doi: 10.1109/AGILE.2014.11. Estimation Based on Story Point and Lines of Code using
[13] R. Popli, N. Chauhan, and H. Sharma, “Prioritising user stories Machine Learning,” Int. J. Comput. Digit. Syst., pp. 1–8, 2021,
in agile environment,” Proc. 2014 Int. Conf. Issues Challenges [Online]. Available:
Intell. Comput. Tech. ICICT 2014, pp. 515–519, 2014, doi: https://journal.uob.edu.bh:443/handle/123456789/4491.
10.1109/ICICICT.2014.6781336. [32] R. K. Mallidi and M. Sharma, “Study on Agile Story Point
[14] M. Usman, J. Börstler, and K. Petersen, “An Effort Estimation Estimation T echniques and Challenges,” Int. J. Comput. Appl.,
Taxonomy for Agile Software Development,” Int. J. Softw. vol. 174, no. 13, pp. 9–14, 2021, doi: 10.5120/ijca2021921014.
Eng. Knowl. Eng., vol. 27, no. 4, pp. 641–674, 2017, doi: [33] L. D. Radu, “Effort prediction in agile software development
10.1142/S0218194017500243. with Bayesian networks,” ICSOFT 2019 - Proc. 14th Int. Conf.
[15] S. Bilgaiyan, S. Mishra, and M. Das, “Effort estimation in Softw. Technol., no. Icsoft, pp. 238–245, 2019, doi:
agile software development using experimental validation of 10.5220/0007842802380245.
neural network models,” Int. J. Inf. Technol., vol. 11, no. 3, pp. [34] H. D. P. Carvalho, M. N. C. A. Lima, W. B. Santos, and R. A.
569–573, 2019, doi: 10.1007/s41870-018-0131-2. de A.Fagunde, “Ensemble Regression Models for Software
[16] A. T . Raslan and N. R. Darwish, “An enhanced framework for Development Effort Estimation: A Comparative Study,” Int. J.
effort estimation of agile projects,” Int. J. Intell. Eng. Syst., Softw. Eng. Appl., vol. 11, no. 3, pp. 71–86, 2020, doi:
vol. 11, no. 3, pp. 205–214, 2018, doi: 10.5121/ijsea.2020.11305.
10.22266/IJIES2018.0630.22. [35] Y. Mahmood, N. Kama, A. Azmi, A. S. Khan, and M. Ali,
[17] A. Saini, L. Ahuja, and S. K. Khatri, “Effort Estimation of “Software effort estimation accuracy prediction of machine
Agile Development using Fuzzy Logic,” 2018 7th Int. Conf. learning techniques: A systematic performance evaluation,”
Reliab. Infocom Technol. Optim. Trends Futur. Dir. ICRITO Softw. - Pract. Exp., 2021, doi: 10.1002/spe.3009.
2018, pp. 779–783, 2018, doi: [36] M. Vyas, A. Bohra, C. S. Lamba, and A. Vyas, “A Review on
10.1109/ICRITO.2018.8748381. Software Cost and Effort Estimation Techniques for Agile
[18] H. Osman and M. Musa, “Asurvey of Agile software Development Process,” Int. J. Recent Res. Asp., vol. 5, no. 1,
estimation methods -IJCST,” Int. J. Comput. Sci. pp. 1–5, 2018.
Telecommun., vol. 7, no. 3, pp. 38–42, 2016, [Online]. [37] M. Arora, S. Chopra, and P. Gupta, “Estimation of regression
Available: http://www.ijcst.org/Volume7/Issue3/p6_7_3.pdf. test effort in agile projects,” Far East J. Electron. Commun.,
[19] M. Miłosz and M. Borys, “Estimation in Agile Projects–Story vol. SpecialVolume3, no. February 2020, pp. 741–753, 2016,
Points Method,” Polish J. Environ. Stud., no. September 2009, doi: 10.17654/ECSV3PII16741.
2009. [38] P. Shakya and S. Shakya, “Critical Success Factor of Agile
[20] C. Prasada Rao, P. Siva Kumar, S. Rama Sree, and J. Devi, An Methodology in Software Industry of Nepal,” J. Inf. Technol.
agile effort estimation based on story points using machine Digit. World, vol. 02, no. 03, pp. 135–143, 2020, doi:
learning techniques, vol. 712. Springer Singapore, 2018. 10.36548/jitdw.2020.3.001.
[21] T ung Khuat and Hanh Le, “ An Effort Estimation Approach
for Agile Software Development using Fireworks Algorithm
Optimized Neural Network.,” Int. J. Comput. Sci. Inf. Secur.,
vol. 14, no. 7, pp. 122–130, 2016.
[22] A. Kaushik, D. K. Tayal, and K. Yadav, “A Comparative
Analysis on Effort Estimation for Agile and Non-agile
Software Projects Using DBN-ALO,” Arab. J. Sci. Eng., vol.
45, no. 4, pp. 2605–2618, 2020, doi: 10.1007/s13369-019-
04250-6.

978-1-6654-9707-7/22/$31.00 ©2022 IEEE 422

Authorized licensed use limited to: BEIJING INSTITUTE OF TECHNOLOGY. Downloaded on October 31,2024 at 04:23:43 UTC from IEEE Xplore. Restrictions apply.