Machine Learning-Based Software Defect Prediction Using Learning to Rank and Linear Regression

Abstract

Software defect prediction plays a crucial role in enhancing software quality and reliability. Predicting defect-prone modules before testing enables efficient resource allocation and reduces maintenance costs. Traditional methods using static code metrics or manual inspection often fail to generalize across diverse datasets and software projects. To address this challenge, this research proposes a hybrid model that integrates Learning to Rank (LTR) algorithms and Linear Regression techniques for effective defect prediction. The model ranks software modules based on their defect likelihood while estimating the defect density using regression analysis. The hybrid approach improves ranking accuracy, prioritization, and interpretability, offering a robust framework for software quality assurance.