Comparative Analysis of Machine Learning Models for Enhanced Software Fault Detection: Performance and Accuracy Evaluation

Abstract

Software defect prediction expectation is a critical portion of program designing that tries to discover and anticipate bugs or imperfections in program frameworks some time recently they happen. The objective is to come up with strategies and models that offer assistance program advancement groups choose which testing assignments are most imperative and how to best utilize their assets. A few machine learning (ML) strategies are utilized to do this. To create forecasts, these frameworks utilize a parcel of distinctive information, like code measures, past imperfection information, and data approximately the coder. The reason of this study is to use different machine learning strategies to make a determining demonstrate for finding computer program bugs. The proposed demonstrate is utilized in tests on the KC2 dataset from NASA's Guarantee library as portion of the study. It was attempted and compared how well diverse machine learning strategies worked. This consider utilized the KC2 dataset and found that the Decision Tree strategy got 73.28% accuracy, Naïve Bayes got 85.96%, K-Nearest Neighbour (KNN) got 82.57%, Support Vector Machine (SVM) got 86.74%, and Random Forest got 84.20%. The study about appeared that these different models can precisely anticipate program bugs when the Guarantee dataset KC2 is utilized. These comes about appear that machine learning models can offer assistance anticipate bugs superior, which can offer assistance groups speed up their tests and make way better utilize of their assets. The study appears that employing a run of machine learning models can enormously move forward the precision of foreseeing computer program abandons. This gives us valuable data almost regions of program frameworks that are more likely to have bugs, which helps keep software quality high.