Software Bug Report Detection Methods Based on Machine Learning Techniques

Software defect prediction, an integral facet of software engineering, proactively identifies and resolves potential flaws before they escalate into disruptive issues during production. Insights gleaned from dedicated research serve as solid support for development teams, pinpointing vulnerable areas within their codebase and optimizing resource allocation. This strategic approach significantly reduces overall expenditure on defect rectification. This paper conducts a comprehensive study, meticulously examining various machine learning algorithms in this domain. It intricately dissects deployment, functionality, and the comparative effectiveness of models like Logistic Regression, K-Nearest Neighbours, Naive-Bayes Classifier, Random Forest Classifier, and Recurrent Neural Networks. This thorough exploration enhances understanding of defect prediction methodologies. It provides a detailed roadmap for optimal strategies, reinforcing software quality measures and mitigating the impact of potential coding irregularities on system integrity and reliability. According to the experimental findings, RNN outperformed the baseline methods in terms of accuracy, F1 score, recall, and precision when identifying software bug reports.