Performance Analysis of Machine Learning Algorithms Using Bagging Ensemble Technique for Software Fault Prediction

Software development comes with a lot of challenges. Developers face various issues with performance and bugs. These issues increase with the scale of the project and if fewer individuals work on the development. It has become necessary to fix various bugs during development to ensure better performance and reduce the chances of failure during the deployment of the software. As a result of this, faults in the software must be predicted during the earlier stages of development. This would help in reducing the cost of maintenance of the software post-deployment. Multiple software fault prediction (SFP) approaches have been proposed to tackle this problem. These approaches can be improved by implementing ensemble techniques. In this paper, we study the effect of the bagging technique and how it helps to improve the predictive capability across various datasets. These datasets are provided and made open source by NASA. Decision Tree Classifier (DTC), Logistic Regression (LR), K-Nearest Neighbors (KNN), Gaussian Naive Bayes (GNB), and Support Vector Classifier (SVC) were used as base classifiers for the bagging method. Random forest (RFC) is an ensemble learning algorithm that uses the bagging technique. Based on the outcome of the results, it was concluded that RFC was the best-performing algorithm.