Leveraging Big Data Analytics for Efficient Bug Localization in Large- Scale Software Projects

Abstract

The exponential growth of software complexity has necessitated more sophisticated techniques for bug localization, which is crucial for maintaining the reliability and efficiency of large-scale software projects. This research paper introduces a novel hybrid approach that synergistically combines Random Forest algorithms with text mining techniques to enhance the accuracy and efficiency of bug localization. By leveraging the strengths of machine learning and natural language processing, our methodology effectively processes and analyzes both structured and unstructured data from software repositories. We present a detailed mathematical framework outlining the integration of these techniques and evaluate the model's performance using precision, recall, and F1- score metrics. Our results demonstrate that while the model shows high precision in identifying non-bug instances, it struggles with accurately detecting bug instances, indicating a need for further refinement of the feature set. The visual analysis of the simulated data highlights the nuanced relationship between code changes and bug occurrences, suggesting that additional context-aware features may improve model performance. The findings emphasize the potential of combining various data analytics techniques for bug localization and point toward future enhancements in feature engineering and model optimization.