Systematic Literature Review: Machine Learning for Software Fault Prediction

This article presents a systematic review of the literature on the use of machine learning for software fault prediction. The objective of the paper is to determine how machine learning algorithms have been used in the approach of models for this type of prediction. The analysis carried out contemplates 52 articles that were published between 2009 and 2022. The study covers the categorization of the algorithms based on the way they were used in the applications. The results showed that the most used algorithms are based on supervised learning, Support Vector Machine (SVM), Random Forest and Naive Bayes; however, the most effective prediction models used a combination of different algorithms.