A Comparative Study of Machine Learning Algorithms for Defect Prediction in Software Systems

Abstract

In the software development lifecycle, defect prediction plays a critical role in ensuring the reliability and performance of software systems. Machine Learning (ML) techniques have gained prominence in recent years for their ability to forecast potential defects in software early, enabling timely interventions and cost reductions. This paper presents a comparative study of various ML algorithms applied to defect prediction in software systems. The study focuses on the efficiency, accuracy, and applicability of common machine learning algorithms, including decision trees, random forests, support vector machines, and deep learning models. Through a comprehensive review of existing literature and empirical analysis, this research aims to evaluate the strengths and weaknesses of these algorithms in real-world defect prediction tasks. The performance of each model is assessed using multiple software datasets, including open-source and industry-specific applications. Metrics such as accuracy, precision, recall, F1 score, and computational cost are used to evaluate model effectiveness. This study also highlights the importance of feature selection, data preprocessing, and model interpretability in enhancing prediction performance. The results of this comparison will provide valuable insights for software developers and researchers aiming to select the most appropriate ML model for defect prediction tasks. Finally, the paper suggests avenues for future research in this domain, focusing on hybrid approaches and ensemble models to further improve prediction accuracy and robustness.

DOI - https://doi.org/10.65525/SVUP.9788199565418.2026.172-177