Abstract:

Software defect prediction plays a crucial role in enhancing software quality and reducing development costs by identifying error-prone modules early in the development life cycle. Traditional defect prediction models often suffer from low accuracy due to imbalanced datasets, irrelevant features, and the inability to generalize across different projects. This study proposes a novel machine learning-based approach to improve defect prediction accuracy by integrating advanced preprocessing techniques, feature selection methods, and ensemble learning algorithms. The proposed model utilizes Synthetic Minority Over-sampling Technique (SMOTE) to address class imbalance, Principal Component Analysis (PCA) for dimensionality reduction, and a hybrid ensemble model combining Random Forest, Gradient Boosting, and Support Vector Machines to enhance prediction performance. The approach is evaluated using publicly available software defect datasets such as NASA MDP and PROMISE repository. Experimental results demonstrate a significant improvement in accuracy, precision, recall, and F1-score compared to traditional methods, indicating the potential of the proposed approach in real-world software engineering applications.

Description:

.