Optimization Accurcay of Predicting Software Defects using Ensemble Machine Learning

Abstract

Software defect prediction plays a vital role in improving software reliability and reducing development costs by identifying fault-prone modules before deployment. Traditional machine learning techniques such as Decision Trees, Support Vector Machines (SVM), and Naïve Bayes have been widely employed for defect prediction; however, their performance often varies due to the imbalance and complexity of software datasets. This study proposes an ensemble machine learning-based approach to enhance the prediction accuracy and robustness of software defect detection. Ensemble techniques such as Random Forest, Gradient Boosting, AdaBoost, and Voting Classifier are analyzed and optimized using hyperparameter tuning strategies to achieve the best predictive performance. Feature selection techniques are also integrated to minimize noise and computational complexity. Publicly available datasets such as NASA and PROMISE repositories are used for experimental validation. Performance metrics such as accuracy, precision, recall, F1-score, and AUC are used to evaluate model efficiency. The results demonstrate that ensemble learning models significantly outperform individual classifiers in terms of both accuracy and generalization. This research highlights the importance of hybrid ensemble approaches in achieving optimized defect prediction accuracy, contributing to reliable and cost-effective software development processes.