A HYBRID MACHINE LEARNING APPROACH FOR SOFTWARE FAULT PREDICTION USING SOFTWARE METRICS

Authors

  • S. S. Prasanna Vengatesan, Dr.N.Balajiraja

Keywords:

Software Fault Prediction, Software Metrics, Hybrid Algorithm, Machine Learning Classifiers, Software Reliability, Defect Detection, Software Quality Assurance

Abstract

Software fault prediction is a critical aspect of software engineering that aims to identify potential defects before they occur, thereby enhancing the reliability and quality of software systems. This study presents an approach for software fault prediction using software metrics derived from McCabe’s and Halstead’s metrics, supplemented by a goal metric. We analyze various datasets, including CM1, KC1, KC2, and PC1, employing classifiers such as Naive Bayes, J48, K-Star, and Random Forest, utilizing MATLAB for implementation. The datasets represent a blend of structural and object-oriented programming, with CM1 and PC1 written in C and KC1 and KC2 in C++. Our evaluation focuses on key performance indicators including accuracy, true positive rate, false positive rate, precision, recall, and F-measure, as well as the area under the ROC curve for a comprehensive assessment. The results demonstrate that the proposed hybrid algorithm effectively predicts faults in both small and large datasets, offering a robust solution for improving software quality. This approach not only streamlines the identification of fault-prone areas but also enables proactive measures to mitigate risks, ultimately leading to more reliable software systems.

Downloads

Published

2024-10-01

How to Cite

S. S. Prasanna Vengatesan, Dr.N.Balajiraja. (2024). A HYBRID MACHINE LEARNING APPROACH FOR SOFTWARE FAULT PREDICTION USING SOFTWARE METRICS. Utilitas Mathematica, 121, 217–228. Retrieved from https://utilitasmathematica.com/index.php/Index/article/view/2045

Citation Check