Litcius/Paper detail

Heterogeneous Stacked Ensemble Classifier for Software Defect Prediction

Somya Goyal

202028 citationsDOI

Abstract

Software defect prediction (SDP) is vital to enhance the software quality with reduced testing cost. It stresses to put more testing efforts on those modules which are susceptible to defects. Hence, the focused testing saves resources and increases the probability to deliver failure-free software product exhibiting desired quality levels. One major hinderance in accurate performance of SDP classifiers is class imbalance. This paper proposes to use stacked ensemble to deal with the class imbalance and enhance the performance of SDP classifiers. The experimental study utilizes the publicly available NASA corpus. The study makes a statistical comparison between the proposed model and baseline models. The results show that the proposed model performs better than baseline models over area under the curve (AUC) and accuracy evaluation metrics.

Topics & Concepts

Computer scienceClassifier (UML)SoftwareMachine learningSoftware qualityBaseline (sea)Artificial intelligenceData miningSoftware bugReliability engineeringClass (philosophy)Software developmentEngineeringProgramming languageGeologyOceanographySoftware Engineering ResearchImbalanced Data Classification TechniquesSoftware Reliability and Analysis Research
Heterogeneous Stacked Ensemble Classifier for Software Defect Prediction | Litcius