Optimizing Bug Resolution: A Data-Driven Developer Recommendation System

Abstract

To deliver a quality project on time during software maintenance, selecting the most suitable developer to assign a newly reported bug is a complex task. The proposed Data-Driven Developer Recommendation System (DDRS) examines developer performance metrics based on bug reports. It recommends using machine learning techniques such as Random Forest, Decision Tree, and Support Vector Machine (SVM). The dataset from four open-source projects (SWT, Eclipse UI, BIRT, and JDT) is initially preprocessed. The algorithm computes severity and priority scores, average bug resolution time, and developer effort before merging these parameters into a Developer Weighted Score for ranking. Machine learning models are trained on the complete dataset with textual features converted to numerical representations via TF-IDF. These algorithms predict developer appropriateness for new bugs by combining predictions and precomputed scores to produce a top ten list. When tested using 10-fold cross-validation, the model displayed better accuracy, scoring up to 98.85% on BIRT, 97.7% on JDT, 91.5% on Eclipse UI, and 96.8% on SWT. The study emphasizes the importance of bug priority, resolution time, workload, and severity in triage. The suggested methodology successfully automates developer assignment, resulting in more efficient and accurate defect resolution in software development projects.