SDP-ML: An Automated Approach of Software Defect Prediction employing Machine Learning Techniques

Software Defect Prediction (SDP) method plays a vital role to ensure the software quality by predicting bugs in software development phase. In addition, this technique also assists developers to minimize the maintenance costs. In this paper, we proposed a model as SDP-ML that utilizes machine learning classification techniques to predict the faults in the software. In particular, the model uses three gradient boosting classification frameworks LightGBM (LGB), XGBoost (XGB), CatBoost (CB) for the prediction. The classification was performed on five publically available NASA Promise datasets, viz., CM1, JM1, PC1, KC1, KC2, and validated using ten-fold cross-validation techniques. Moreover, the remarkable evaluation techniques Precision, Recall, F1-score, and Accuracy were employed to evaluate the results. The outcomes demonstrated the dominant performance of LightGBM with GridSearchCV library (98%) than other algorithms considering the average F1-score. However, analysts can achieve free comprehension from this examination while choosing an automated research field for their planned application.