Research Progress of Machine Learning in Software Defect Prediction

Abstract

Software defect prediction is of great significance to ensuring software security. Appropriate metrics are an effective way to predict whether a program contains software flaws. In recent years, machine learning techniques have been widely used to measure programs to predict software defects and have achieved good results. In order to clarify the research progress of software defect prediction based on machine learning, this paper reviews the research results in this field in the past three years from two aspects: within-project defect prediction and cross-project defect prediction. Firstly, the software defect prediction framework based on machine learning and its development are introduced. Secondly, this paper summarizes the research results in the field of defect prediction in the past three years from two aspects: within-project defect prediction and cross-project defect prediction. Finally, based on the research results so far, the status and hot spots of machine learning in software defect prediction are discussed, which provides possible help for the development direction of machine learning in software defect prediction.