Research on Cross-Project Software Defect Prediction Based on Machine Learning

Abstract

In recent years, machine learning technology has developed vigorously. The research on software defect prediction in the field of software engineering is increasingly adopting various algorithms of machine learning. This article has carried out a systematic literature review on the field of defect prediction. First, this article studies the development process of defect prediction, from correlation to prediction model. then this article studies the development process of cross-project defect prediction based on machine learning algorithms (naive Bayes, decision tree, random forest, neural network, etc.). Finally, this paper looks forward to the research difficulties and future directions of software defect prediction, such as imbalance in classification, cost of data labeling, and cross-project data distribution.