Evaluation of Classification Algorithms for Software Defect Prediction

Abstract

For Software Engineering industry is essential to build software with high quality, but this has become difficult these days with the fact that size and complexity of developed software is high. For this reason, it is important to predict the quality of software in early phases, by this way, it helps to reduce testing resources. Some classification algorithms have been used for software defect prediction. In this paper five algorithms were evaluated and compared according to accuracy metric using three PROMISE datasets. For the evaluation, two classifiers were used which are focus on decision trees and three Bayesian classifiers. Results shows that even when percentages of accuracy are high on each classifier, decision trees got the highest percentage compared to Bayesian classifiers.