Software Defect Prediction Approaches: A Review

Software testing is a time-consuming and costly task, as it involves testing all software modules. To minimize the cost and effort of software testing, automatic defect detection can be used to identify the defective modules during the early stages. These aid software testers in detecting the modules that require intensive testing. Therefore, automatically predicting software defects has become a critical factor in software engineering. This paper explores the existing methods and techniques on software defect prediction (SDP) and lists the most popular datasets that are used as benchmarks in SDP. In addition, it discusses the approaches to overcome the class imbalance problem, which usually occurs in the benchmark datasets for SDP problems. This paper can be helpful for researchers in software engineering and other related areas.