An Enhanced Ensemble-Based Software Defect Prediction Framework Using Software Metrics and Machine Learning Techniques
International Journal of Advanced Research in Science Communication and Technology · 25.06.2026 · 426
DOI: 10.48175/ijarsct-37155 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Advanced Research in Science Communication and Technology, online erschienen am 25.06.2026. Verfasst von Rajeev P R and Sruthi R Varrier. Die Arbeit nennt 53 Literaturangaben.
Abstract (Originalsprache)
Software defect prediction plays a vital role in software quality assurance by enabling the early identification of defect-prone software modules and facilitating efficient allocation of testing resources. The increasing complexity of modern software systems necessitates the development of accurate and reliable defect prediction techniques to reduce maintenance costs and improve software reliability. Traditional software defect prediction approaches often suffer from limitations related to noisy datasets, redundant software metrics, and the inability of individual classifiers to effectively capture complex relationships among software attributes. To address these challenges, this paper proposes an Enhanced Ensemble Software Defect Prediction (EESDP) framework that integrates software metrics, data preprocessing, feature ranking, and ensemble machine learning techniques for effective defect prediction. The proposed framework employs data cleaning and preprocessing mechanisms to improve dataset quality, followed by feature ranking to identify significant software metrics. Ensemble classification is achieved through the integration of Random Forest and Gradient Boosting classifiers using a voting-based decision mechanism. The framework is evaluated using benchmark NASA Metrics Data Program (MDP) datasets, including KC1, KC2, CM1, PC1, and JM1. The proposed approach aims to improve prediction robustness by combining the strengths of multiple classifiers while reducing the influence of noisy and redundant attributes. The effectiveness of the proposed framework is discussed using standard performance measures such as Accuracy, Precision, Recall, and F1-Score. The framework provides a foundation for future experimental validation using benchmark software defect prediction datasets. The proposed EESDP framework provides a practical and scalable solution for software defect prediction and contributes to the development of reliable software systems..
Zitieren
APA
Varrier, R. P. R. A. S. R. (2026). An Enhanced Ensemble-Based Software Defect Prediction Framework Using Software Metrics and Machine Learning Techniques. International Journal of Advanced Research in Science Communication and Technology, 426. https://doi.org/10.48175/ijarsct-37155
Vancouver
Varrier RPRASR. An Enhanced Ensemble-Based Software Defect Prediction Framework Using Software Metrics and Machine Learning Techniques. International Journal of Advanced Research in Science Communication and Technology. 2026:426. doi:10.48175/ijarsct-37155
BibTeX
@article{Varrier2026,
author = {Varrier, Rajeev P R and Sruthi R},
title = {An Enhanced Ensemble-Based Software Defect Prediction Framework Using Software Metrics and Machine Learning Techniques},
journal = {International Journal of Advanced Research in Science Communication and Technology},
year = {2026},
pages = {426},
doi = {10.48175/ijarsct-37155}
}
EUVIMED