Modeling of software project effort estimation: a comparative performance evaluation of optimized soft computing-based methods

Software cost estimation has always been a significant problem for software development teams and should be considered during the early stages of a software project development process. Inadequate knowledge about final requirements and the presence of imprecise and vague requirements are among the p...

Full description

Saved in:
Bibliographic Details
Published in:International journal of information technology (Singapore. Online) 2022, Vol.14 (5), p.2487-2496
Main Authors: Sharma, Sudhir, Vijayvargiya, Shripal
Format: Article
Language:English
Subjects:
Artificial Intelligence
Computer Imaging
Computer Science
Image Processing and Computer Vision
Machine Learning
Original Research
Pattern Recognition and Graphics
Software Engineering
Vision
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Software cost estimation has always been a significant problem for software development teams and should be considered during the early stages of a software project development process. Inadequate knowledge about final requirements and the presence of imprecise and vague requirements are among the primary causes of unreliable estimations in this field. Though several techniques and models for effort and cost estimation have been proposed in recent years, improving their accuracy has always been a contentious topic, and researchers' efforts in this field continue. This study aims to compare and evaluate the effort estimation performance of four distinct approaches: Localized Neighbourhood Mutual Information based neural network (LNI-NN), Neuro-fuzzy logic (NFL), Adaptive GA-based neural network (AGANN), and GEHO-based NFN (GEHO-NFN). All four modeling approaches were tested & validated on five distinct datasets, namely cocomo81, cocomonasa1, cocomonasa v2, desharnais, and china datasets selected from the promise database. All the results for software effort estimation were compared, and output was evaluated using performance assessment metrics such as MMRE (Mean magnitude of relative error), MdMRE (Median of Magnitude relative error), RMSE (Root mean square error), and PRED (Prediction accuracy). Results show that GEHO-based NFN (GEHO-NFN) performance is better than other mentioned popular soft computing methods.
ISSN:2511-2104
2511-2112
DOI:10.1007/s41870-022-00962-5