MCOT-KB: Multi-source cross-project defect prediction with optimal transport domain adaptation and KMMBagging

Key Points

• Defect prediction accuracy increased significantly with optimal transport methods, improving cross-project outcomes.
• The study shows a substantial improvement in performance metrics, indicating 15% higher defect detection rates across different datasets.
• Analysis employing multi-source data utilized optimal transport for domain adaptation, integrating diverse project data efficiently.
• Highlights the potential for improved software reliability through advanced defect prediction techniques, yet external validation is essential.