An Ensemble Bootstrap Aggregation for Just-in-Time Defect Prediction on a Code Change Level

Just-in-time defect prediction (JITDP) is a technique for predicting if a code change is defective. JITDP ensures software quality throughout the design phase and requires developers to check and resolve defects on time. Several recent studies have proposed using JITDP to detect changes that potentially create defects at check-in time. JITDP techniques utilize change metrics obtained from software repositories to forecast defect-inducing changes. In this research, we presented Bootstrap aggregation to detect changes that potentially create defects at check-in time by employing the publicly accessible change metrics dataset Mozilla. The proposed method has been compared to several machine learning techniques to verify the proposed approach's performance. The findings reveal that the proposed approach beat all comparable machine learning methods in various classification measures. In addition to the proposed approach, the decision tree and k nearest neighbor algorithms performed well.