Accuracy Prediction of Ensemble Deep Learning Model through Software Defect Prediction

Defect detection and repair are critical to the performance and dependability of modern software development. To evaluate large software repositories and find errors early in the development process, this study proposes a novel deep learning (DL) method for software defect prediction (SDP). This work aims to methodically ascertain, and evaluate MLP, CNN, and LSTM models independently and finally ensemble it for better SDP. We examine the MLP, CNN, LSTM individually and compare them with the ensemble model. We achieved an accuracy of 0.78, indicating promising results for defect prediction in software development in imbalanced data set.