An empirical study on the effectiveness of data resampling approaches for cross‐project software defect prediction

Cross-project defect prediction (CPDP), where data from different software projects are used to predict defects, has been proposed as a way provide for that lack historical data. Evaluations of CPDP models using the Nearest Neighbour (NN) Filter approach have shown promising results in recent studies. A key challenge with defect-prediction datasets is class imbalance, is, highly skewed non-buggy modules dominate buggy modules. In past, resampling approaches applied within-projects help alleviate negative effects imbalance datasets. To address issue CPDP, authors assess impact on after NN applied. The performance five oversampling (MAHAKIL, SMOTE, Borderline-SMOTE, Random Oversampling and ADASYN) three undersampling (Random Undersampling, Tomek Links One-sided selection) investigated compared without resampling. examined six 34 extracted PROMISE repository. authors' show there significant positive effect performance, suggesting quality teams researchers should consider applying improved recall (pd) g-measure performance. However, if goal improve precision reduce false alarm (pf) then be avoided.