A Hybrid Machine Learning Framework Using Random Forest and XGBoost for Software Bug Prediction

  • Unique Paper ID: 175775
  • Volume: 11
  • Issue: 11
  • PageNo: 4664-4667
  • Abstract:
  • Code smells, indicating poor design or implementation choices, can harm software maintainability and increase bug-proneness. This study explores the significance of code smell metrics in prediction models for detecting bug-prone code modules. By incorporating smell-based metrics, we aim to enhance bug prediction accuracy. Using 14 open-source projects from the PROMISE repository, all written in Java, we trained models with metrics like F1-score, accuracy, precision, and recall. Classifiers like Naïve Bayes, Random Forest (RF), Support Vector Machine (SVM), Logistic Regression, and k-Nearest Neighbor were applied. RF and SVM outperformed the other methods, delivering higher accuracy both within versions and across projects, proving their effectiveness in predicting buggy components.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 11
  • Issue: 11
  • PageNo: 4664-4667

A Hybrid Machine Learning Framework Using Random Forest and XGBoost for Software Bug Prediction

Related Articles