BIO-INSPIRATION BASED FUZZY RULE OPTIMIZATION FOR SOFTWARE DEFECT PREDICTION

Abstract

There are numerous data mining algorithms available to perform the software defect prediction task. The prediction of the software defect is used to improve the quality of the software and the efficiency of its maintenance system. Software defect prediction using classification algorithms was supported by many researchers. The issue related to such classifiers is that they generate more number of rules in which most of them are irrelevant and due to it there is a chance of high degree of misclassification on data instances. This leads to the necessity of using rule optimization technique which will strongly control the population of rules generated by the classifier by applying optimization approaches to eradicate the irrelevant and redundant rules.  Moreover this rule optimizer makes the classifier to improve classification performance more effectively with less time complexity. This paper is inspired on the bio inspirational approach for applying the optimization technique on the fuzzy classifier thus this proposed work adapted cuckoo search method to find the best rules and pruning the irrelevant rules produced by the fuzzy inference engine. The simulation result proves the performance of the proposed Cuckoo search based Fuzzy Rule optimizer in software Defect Prediction (CSFZSDP) in an effective manner.