Enhancing Software Defect Prediction Accuracy using Stacking Ensemble Model

Software defect prediction is the culmination process in software development industries. Software defect prediction using machine learning (ML) is an increasingly vital approach for improving software quality and reducing development costs by anticipating potential defects before they impact the final product. Machine learning models offer a powerful means to predict software defects by analyzing historical data and identifying patterns that indicate defect-prone areas. Different supervised and unsupervised classification models are used to predict software defect at early stage of development life cycle. But they face some challenges due to data quality and model adaptability. Recent advancements address these challenges through techniques such as ensemble learning, which combines predictions from multiple models to enhance accuracy and robustness. In our research, we have compared the predicted results using Support Vector Classifier (SVC), Extreme Gradient Boosting (XGB), Random Forest (RF) and Voting Classifier with our proposed stacking classifier designed with SVC, RF and XGB. In our experimental study, it shows that proposed stacking model outperforms in more accurate detection of software defects for multiple datasets. This research findings can greatly advance the effectiveness of defect prediction, contributing to the development of higher-quality software systems.