Software defect prediction based on geometric mean for subspace learning

Due to the confusion of fault-prone software modules and non-fault-prone ones, and the limit of traditional mothed such as LDA and PCA, the performance of software defect prediction model is difficult to improve. In this paper, we present GMCRF, a method based on dimensionality reduction technique and conditional random field (CRF) for software defect prediction. In our proposed method, firstly, we leverage geometric mean for subspace learning to choose the best combination of features from data set. Secondly, we propose to apply the best combination of features which is selected by geometric mean-based approach in CRF model. Interestingly, we find that the GMCRF method achieves much better final results than the other approach as shown in the software defect data classification task.