SMOTE-Based Homogeneous Ensemble Methods for Software Defect Prediction

Abstract

Class imbalance is a prevalent problem in machine learning which affects the prediction performance of classification algorithms. Software Defect Prediction (SDP) is no exception to this latent problem. Solutions such as data sampling and ensemble methods have been proposed to address the class imbalance problem in SDP. This study proposes a combination of Synthetic Minority Oversampling Technique (SMOTE) and homogeneous ensemble (Bagging and Boosting) methods for predicting software defects. The proposed approach was implemented using Decision Tree (DT) and Bayesian Network (BN) as base classifiers on defects datasets acquired from NASA software corpus. The experimental results showed that the proposed approach outperformed other experimental methods. High accuracy of 86.8% and area under operating receiver characteristics curve value of 0.93% achieved by the proposed technique affirmed its ability to differentiate between the defective and non-defective labels without bias.

Software Defect Prediction Using Ensemble Learning: A Systematic Literature Review

“…The results showed that no single ensemble method outperformed others in all datasets. However, the [72] 2020 Springer Link Conference International Conference on Computational Science and Its Applications (ICCSA 2020) [73] 2021 Springer Link Journal Applied Intelligence [74] 2021 Springer Link Journal Neural Computing and Applications researchers observed that the ensembles of a few ranking techniques performed better than the ensembles of many ranking techniques. In [59], a review of state-of-the-art ensemble techniques for class imbalance problems was conducted.…”

“…Their technique showed promising results with the highest AUC of 0.93 in one group of source and target datasets. In [72], researchers proposed a method using SMOTE and homogeneous ensemble methods (bagging and boosting) to improve the performance of defect prediction models. They employed DT and BN as baseline classifiers in their model.…”

“…[71] proposed a model to predict defect across projects with a HDP [72] proposed a method using SMOTE and homogeneous ensemble methods (bagging and boosting)…”

“…In [71], the AUC, recall, precision and F-measure were used to measure the performance of the proposed ensemble learning technique. In [72], the AUC, Accuracy and F-measure are employed to measure the performance of the proposed defect prediction model. In [73], five performance measures namely precision, recall, AUC (area under ROC curve), specificity, and G-means were used.…”

Detecting Generic Network Intrusion Attacks using Tree-based Machine Learning Methods

“…Decision tree algorithms are a family of machine learning classification and regression algorithms that fits a model on a given dataset having considered the entropy of some or all attributes for making its splitting decision. Tree-based machine learning algorithms are widely used and acceptable for various research and industrial areas, even as distant as software defect prediction in the field of software engineering [24] and even for the prediction of factors in educational management [25]. Decision Tree models are known to always produce interpretable models.…”