An empirical analysis of the statistical learning models for different categories of cross-project defect prediction