Optimization Accuracy of Software Defect Prediction using Feature Selection and Machine Learning Techniques

Abstract

Software defect prediction is an important research area in software engineering that aims to identify defective software modules during the software development lifecycle. The presence of software defects can reduce system reliability, increase maintenance costs, and affect overall software quality. Traditional software testing approaches often require considerable time and resources, making machine learning techniques valuable for automated defect prediction. However, the presence of irrelevant and redundant features, high-dimensional datasets, and class imbalance can negatively affect prediction accuracy. To address these challenges, this study proposes an optimized software defect prediction approach using feature selection and machine learning techniques. The proposed methodology involves collecting historical software datasets, preprocessing the data, selecting relevant software metrics, and applying machine learning algorithms for defect classification. Feature selection techniques are utilized to identify the most significant attributes, reduce data dimensionality, and improve model efficiency. Various machine learning algorithms, including Decision Tree, Random Forest, Support Vector Machine, K-Nearest Neighbors, Logistic Regression, and ensemble learning techniques, can be investigated for predicting defective and non-defective software modules. The performance of the models is evaluated using accuracy, precision, recall, F1-score, sensitivity, and specificity. The study aims to investigate the effectiveness of feature selection in improving classification performance and reducing computational complexity. The proposed approach is expected to support early defect identification, optimize software testing activities, and improve software quality and reliability.