Clown Fish Optimized – Modified Support Vector Machine (CFO-MSVM) for Software Defect Prediction
Year: 2024
Type: article
Abstract: The growing software industry and the necessity for software development has increased rapidly. The biggest challenge is to develop software in minimal time with fewer resources and bug-free. Software defect prediction has the privilege of predicting the software bug at the earliest to avoid chaos. This paper presents a novel method combining the nature-inspired optimization technique with the Support Vector Machine, proposing a Clown Fish Optimized – Modified Support Vector Machine (CFO-MSVM) classifier for earlier effective classification of the bug. The objective function of the proposed classifier is to tune the hyperparameter of SVM following the swarm intelligence of the clown fish crowd. The Java Developers Toolkit (JDT) dataset from AEEEM repository is used as the bench marker to validate the CFO-MSVM classifier. The classifier is investigated using a grid search for the regularization parameter C, and the number of iterations is set to 100. Precision, Recall and F Score Metrics are used for evaluation. FMI and MCC statistical measurements are employed to define accuracy further. The CFO-MSVM classifier segregates the defect and non-defect modules with 87.32 % accuracy compared to the existent SVM and SVM-GA classifiers, which have 50.83% and 65.00%, respectively.
Authors: Medhunhashini D R, Jeen Marseline KS
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4403360424