A Multi-objective Bat Algorithm for Software Defect Prediction

Both the class imbalance of datasets and parameter selection of support vector machine (SVM) play an important role in the process of software defect prediction. To solve these two problems synchronously, the false positive rate (pf) and the probability of detection (pd) are considered as two objective functions to construct the multi-objective software defect prediction model in this paper. Meanwhile, a multi-objective bat algorithm (MOBA) is designed to solve this model. The individual update strategy in the population is performed using the individual update method in the fast triangle flip bat algorithm, and the non-dominated solution set is used to save the better individuals of the non-defective module and the support vector machine parameters. The simulation results show that MOBA can effectively save resource consumption and improve the quality of software compared with other commonly used algorithms.