Classification Algorithms for Software Defect Prediction: A Systematic Literature Review

Abstract

Within Software Engineering, it is essential to build quality software. An obstacle to the after mentioned are the defects that can be found in any phase of software development. That is why the area of software defect prediction emerged, in which different algorithms have already been evaluated, studied, and proposed. The objective of this paper is to carry out a Systematic Literature Review, to know which are the classification algorithms that help to predict software defects. In the same way, it is intended to know the features, metrics and what has been the precision of the classification algorithms for software defect prediction. This paper shows that the most widely used classifiers to predict defects are Naive Bayes and Random Forest, while those that show the best results are Naive Bayes and Boosting. Finally, it is highlighted that Precision and Recall are the most used metrics for model validation.