A Novel Approach To Improve Software Defect Prediction Using Machine Learning

Abstract

One of the hottest topics in software engineering now is defect predictionthe success of the product depends on bridging the gap between software engineering and data mining.Source code mistakes are predicted via software defects prediction before testing.Techniques for forecasting software defects, such as clustering, statistical methods,mixedalgorithms, neural networkbased metrics, black box and white box testing, and machine learning, are frequently used to investigate the effect area in software.The main contribution of this study is the first use of feature selection to increase the accuracy of defect prediction by machine learning classifiers.The goal of this research is to improve defect prediction accuracy in five NASA data sets: PC1, CM1, JM1, KC2, and KC1.These NASA data sets are publicly available.