Software Defect Prediction Technology Based on Fuzzy Support Vector Machine

Software defect prediction(DP) is one of the most active research areas in software engineering. DP models can provide error-prone source code components or changes, etc., so that quality assurance teams can deal with error-prone source code with more effort, effectively allocating limited resources to verify software products. As software projects grow in size, DP techniques will play a vital role in the work of developers, helping them to develop more reliable software products, resulting in faster time to market. In this paper, fuzzy support vector machine (FSVM) is introduced to build a software DP model, which is trained on the data set by the FSVM classification module. In this paper, by comparing the mean square error of the training data of SVM and FSVM, in the subset of low-defect modules, the training result of FSVM is not as good as that of SVM, that is, the prediction accuracy of SVM is higher; while in the subset of modules with high number of defects, the training result of FSVM is better than that of SVM, that is, the prediction accuracy of FSVM is higher.