Reproduce “Interval Regression: A Comparative Study with Proposed Models” 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
Tung L Nguyen, Toby Dylan Hocking
Researchers conducted a comprehensive comparative analysis of existing and proposed interval regression models, introducing adaptations like Maximum Margin Interval Forest (MMIF) and evaluating them across 39 datasets. The study identified MMIF as the top performer overall, providing empirical guidance for selecting models when target values are interval-censored.
Key insights
- No single Interval Regression model consistently outperforms all others across every scenario, highlighting the importance of selecting models based on specific dataset characteristics.
- Ensemble methods, particularly the proposed Maximum Margin Interval Forest (MMIF), generally exhibit superior overall performance and consistency compared to individual models.
- Tree-based models excel at capturing nonlinear relationships, while linear models are effective for datasets with inherently linear patterns.
- The AFT model within XGBoost demonstrated limitations, particularly when handling left-censored intervals, indicating its specific applicability might be restricted in general interval regression contexts.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.