Software Defect Prediction using Feature Selection and Random Forest Algorithm

Abstract

Software testing is the most important task in software production and it takes a lot of time, cost and effort. Thus, we need to reduce these resources. Software Defect Prediction (SDP) mechanisms are used to enhance the work of SQA process through the prediction of defective modules, many approaches have been conducted by researchers in order to predict the fault-proneness modules. This paper proposed an approach for the SDP purpose, it employs two existed algorithms to have a high performance, that are the Bat-based search Algorithm (BA) for the feature selection process, and the Random Forest algorithm (RF) for the prediction purpose. This paper also has tested a number of feature selection algorithms and classifiers to see their effectiveness in this problem.