An End-to-End S-AF-RCNN based Framework for Just-in-Time Defect Prediction

The problem of defect prediction is very important, especially at the change-level. Change-level defect prediction, also known as just-in-time defect prediction, has the potential to not only guarantee software quality during the development process but also force the developers to find and correct errors as soon as possible. Deep learning is a current hot issue in the literature on machine learning. It has not yet been determined whether deep learning can be utilized to enhance the fault prediction performance. Preprocessing in the suggested method makes use of independence analysis and redundancy analysis. It uses the SMOTE algorithm to choose features. The proposed model is trained with S-AF-RCNN. When compared to more conventional approaches like RCNN and Fast- RCNN, the proposed method performs admirably.