24 Using Clustering Techniques Goes with Genetic Algorithm to Improve Software Defect Prediction Available to Purchase

Software defect prevention is an important activity to improve software quality, in which the possible defects can be predicted and avoided in advance. Identifying the actions that may cause defect before execution is difficult. To settle this problem, this study proposes a defect prediction approach, in which the data of defects and actions are collected to construct the prediction models. The proposed approach is adapted from Association Rule based Defect Prediction (ARDP), and applies Density-Based clustering algorithm with Genetic algorithm (CBDPGA) on the collected data. The improvements of the proposed approach include decreasing the impact of outlier to defect prediction models, and increasing the performance of prediction model. To demonstrate the performance of this study, this approach is applied on the same dataset used in Association Rule based Defect Prediction (ARDP) approach. The results show that the proposed approach can be used to improve prediction performance.