Decision-Level Defect Prediction Based on Double Focuses

This research mainly expounds upon the decision-level software defect prediction theory. The defect characteristics is the first research focus. For the first research focus, a characteristic comparison set is built out of the existing defect characteristics according to the dissimilarity of defect characteristics and the defect characteristics are organized outside the characteristic comparison set into some defect characteristic clusters to reduce the scale of the characteristic data. The defects is the second research focus. For the second research focus, the vector weights are assigned to the defect characteristics contained in the defects according to the minimum critical characteristic set. Moreover, the multi-agent algorithm integration technology is used to predict defects according to the repulsive relationship between similar defect clusters.