Reproduce “The Integrity of Machine Learning Algorithms against Software Defect Prediction” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Param Khakhar and, Rahul Kumar Dubey
This paper evaluates combining Cluster-based Over-Sampling with Noise Filtering (KMFOS) and the Online Sequential Extreme Learning Machine (OS-ELM) for software defect prediction. This approach significantly improved recall scores by 40-70 percentage points across various NASA MDP datasets while maintaining computational efficiency, effectively addressing the challenge of class imbalance.
Key insights
- Class imbalance is a critical barrier to effective software defect prediction, and addressing it is essential for acceptable model performance.
- KMFOS effectively mitigates class imbalance by generating synthetic minority samples while filtering noise, preserving data distribution.
- OS-ELM, when combined with appropriate oversampling, is a viable and computationally efficient classifier for SDP, offering stable performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.