Intelligent Bug Management: Hypertuned Supervised Classification for Automatic Bug Report Prioritization

In order to effectively prevent software failures, finding and prioritizing bugs are crucial. The quantity and complexity of bugs can increase with the size and complexity of a software product. Software testers typically generate bug reports to systematically handle defects, which contain flaws found during testing. Bug fixing has become a critical and costly operation in the software development life cycle due to the exponential rise of bugs in software products. Manually prioritizing bugs in a timely and accurate manner can impede the handling of critical bugs. The time required to handle incoming bug reports can be minimized through automated priority prediction processes. Existing machine learning-based automated bug prioritization models have not achieved significant accuracy. Therefore, to achieve significant prioritization accuracy, we have developed a machine learning-based model incorporating hyperparameter tuning methods. The selected dataset from the Chromium repository encompasses information on six unique types of issues, encompassing bug, feature, and others. These data points are distributed across 1693 components such as “UI”, “BROWSER”, and “GetSetUp” and are classified into eight levels of priority. This showcases the model’s ability to efficiently prioritize bug reports. Hyperparameter tuning optimizes model parameters, enhancing accuracy by adapting to bug report complexity, leading to significant improvements in bug prioritization precision. Utilizing over 12,000 bug reports, our model demonstrates superior performance with an accuracy rate of 96.77%.