Software Defect Prediction: A Comparative Analysis of Machine Learning Techniques

Abstract

Listen

Translate article

The early prediction of defective modules in developing software can help the development team to utilize the available resources efficiently to deliver high quality software product in limited time. The classification technique, which works by extracting hidden defective patterns among software properties, is a good way to determine errors. In this research, twelve commonly used NASA datasets are utilised to forecast software flaws using a variety of machine learning classification methods. Nave Bayes (NB), Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), AdaBoost Algorithm(ADB), Gradient boosting(GB), Logistic Regression(LR), XG-Boost Algorithm Multi - layer perceptron (MLP), K Nearest Neighbor (KNN) are some of the classification techniques. Performance of used classification techniques is evaluated by using various measures such as: Precision, Recall, F-Measure, Accuracy and Overall average Performance measure. Results shows that the Mean Gradient boosting as well as Logistic Regression provide greater performance for Defect Prediction among other Machine Learning Classifiers. Furthermore, the experimental results reveal that Gradient Boost(GB) outperforms all nine machine learning classifiers in terms of overall average performance.