Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms

In today’s fast-paced world of software development, it is essential to ensure that programs run smoothly without any issues. When dealing with complex applications, the objective is to predict and resolve problems before they escalate. The prediction of software defects is a crucial element in main...

Full description

Saved in:
Bibliographic Details
Main Authors: Ayesha Siddika, Momotaz Begum, Fahmid Al Farid, Jia Uddin, Hezerul Abdul Karim
Format: Article
Language:English
Published: MDPI AG 2025-07-01
Series:Eng
Subjects:
Online Access:https://www.mdpi.com/2673-4117/6/7/161
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849418562841608192
author Ayesha Siddika
Momotaz Begum
Fahmid Al Farid
Jia Uddin
Hezerul Abdul Karim
author_facet Ayesha Siddika
Momotaz Begum
Fahmid Al Farid
Jia Uddin
Hezerul Abdul Karim
author_sort Ayesha Siddika
collection DOAJ
description In today’s fast-paced world of software development, it is essential to ensure that programs run smoothly without any issues. When dealing with complex applications, the objective is to predict and resolve problems before they escalate. The prediction of software defects is a crucial element in maintaining the stability and reliability of software systems. This research addresses this need by combining advanced techniques (ensemble techniques) with seventeen machine learning algorithms for predicting software defects, categorised into three types: semi-supervised, self-supervised, and supervised. In supervised learning, we mainly experimented with several algorithms, including random forest, k-nearest neighbors, support vector machines, logistic regression, gradient boosting, AdaBoost classifier, quadratic discriminant analysis, Gaussian training, decision tree, passive aggressive, and ridge classifier. In semi-supervised learning, we tested are autoencoders, semi-supervised support vector machines, and generative adversarial networks. For self-supervised learning, we utilized are autoencoder, simple framework for contrastive learning of representations, and bootstrap your own latent. After comparing the performance of each machine learning algorithm, we identified the most effective one. Among these, the gradient boosting AdaBoost classifier demonstrated superior performance based on an accuracy of 90%, closely followed by the AdaBoost classifier at 89%. Finally, we applied ensemble methods to predict software defects, leveraging the collective strengths of these diverse approaches. This enables software developers to significantly enhance defect prediction accuracy, thereby improving overall system robustness and reliability.
format Article
id doaj-art-aec7c1212d2c460586f2486ae69141db
institution Kabale University
issn 2673-4117
language English
publishDate 2025-07-01
publisher MDPI AG
record_format Article
series Eng
spelling doaj-art-aec7c1212d2c460586f2486ae69141db2025-08-20T03:32:26ZengMDPI AGEng2673-41172025-07-016716110.3390/eng6070161Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning ParadigmsAyesha Siddika0Momotaz Begum1Fahmid Al Farid2Jia Uddin3Hezerul Abdul Karim4Department of Computer Science and Engineering, Dhaka University of Engineering & Technology, Gazipur 1707, BangladeshDepartment of Computer Science and Engineering, Dhaka University of Engineering & Technology, Gazipur 1707, BangladeshCentre for Image and Vision Computing (CIVC), COE for Artificial Intelligence, Faculty of Artificial Intelligence and Engineering (FAIE), Multimedia University, Cyberjaya 63100, Selangor, MalaysiaArtificial Intelligence and Big Data Department, Endicott College, Woosong University, Daejeon 34600, Republic of KoreaCentre for Image and Vision Computing (CIVC), COE for Artificial Intelligence, Faculty of Artificial Intelligence and Engineering (FAIE), Multimedia University, Cyberjaya 63100, Selangor, MalaysiaIn today’s fast-paced world of software development, it is essential to ensure that programs run smoothly without any issues. When dealing with complex applications, the objective is to predict and resolve problems before they escalate. The prediction of software defects is a crucial element in maintaining the stability and reliability of software systems. This research addresses this need by combining advanced techniques (ensemble techniques) with seventeen machine learning algorithms for predicting software defects, categorised into three types: semi-supervised, self-supervised, and supervised. In supervised learning, we mainly experimented with several algorithms, including random forest, k-nearest neighbors, support vector machines, logistic regression, gradient boosting, AdaBoost classifier, quadratic discriminant analysis, Gaussian training, decision tree, passive aggressive, and ridge classifier. In semi-supervised learning, we tested are autoencoders, semi-supervised support vector machines, and generative adversarial networks. For self-supervised learning, we utilized are autoencoder, simple framework for contrastive learning of representations, and bootstrap your own latent. After comparing the performance of each machine learning algorithm, we identified the most effective one. Among these, the gradient boosting AdaBoost classifier demonstrated superior performance based on an accuracy of 90%, closely followed by the AdaBoost classifier at 89%. Finally, we applied ensemble methods to predict software defects, leveraging the collective strengths of these diverse approaches. This enables software developers to significantly enhance defect prediction accuracy, thereby improving overall system robustness and reliability.https://www.mdpi.com/2673-4117/6/7/161software defectsemi-supervised learningself-supervised learningsupervised learningensemblemachine learning
spellingShingle Ayesha Siddika
Momotaz Begum
Fahmid Al Farid
Jia Uddin
Hezerul Abdul Karim
Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms
Eng
software defect
semi-supervised learning
self-supervised learning
supervised learning
ensemble
machine learning
title Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms
title_full Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms
title_fullStr Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms
title_full_unstemmed Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms
title_short Enhancing Software Defect Prediction Using Ensemble Techniques and Diverse Machine Learning Paradigms
title_sort enhancing software defect prediction using ensemble techniques and diverse machine learning paradigms
topic software defect
semi-supervised learning
self-supervised learning
supervised learning
ensemble
machine learning
url https://www.mdpi.com/2673-4117/6/7/161
work_keys_str_mv AT ayeshasiddika enhancingsoftwaredefectpredictionusingensembletechniquesanddiversemachinelearningparadigms
AT momotazbegum enhancingsoftwaredefectpredictionusingensembletechniquesanddiversemachinelearningparadigms
AT fahmidalfarid enhancingsoftwaredefectpredictionusingensembletechniquesanddiversemachinelearningparadigms
AT jiauddin enhancingsoftwaredefectpredictionusingensembletechniquesanddiversemachinelearningparadigms
AT hezerulabdulkarim enhancingsoftwaredefectpredictionusingensembletechniquesanddiversemachinelearningparadigms