Optimal Feature Selection through Search-Based Optimizer in Cross Project

Cross project defect prediction (CPDP) is a key method for estimating defect-prone modules of software products. CPDP tempting approach since it provides information about predicted defects those projects in which data are insufficient. Recent studies specifically include instructions on how to pick training from large datasets using feature selection (FS) process contributes the most end results. The classifier helps classify picked-up dataset specified classes order predict defective and non-defective classes. aim our research select optimal set features multi-class through search-based optimizer CPDP. We used explanatory type quantitative experimentation. have F1 measure as dependent variable while independent variables we KNN filter, ANN random forest ensemble (RFE) model, genetic algorithm (GA), classifiers manipulative variables. Our experiment follows 1 factor treatment (1F1T) RQ1 whereas RQ2, RQ3, RQ4, there 2 treatments (1F2T) design. first carried out analysis (EDA) know nature dataset. Then pre-processed by removing solving issues identified. During preprocessing, analyze that data; therefore, rank multiple sets info gain get maximum variation To remove noise, use ANN-filter significant results more than 40% 60% compared NN filter with base paper (all, ckloc, IG). applied i.e., best model 30% 50% instance (GIS). compare F1-measure almost 35% paper. validate Wilcoxon Cohen’s d test.