Research map / Learning theory and algorithms / Prediction and decision losses
A Unified Framework for Random Forest Prediction Error Estimation
Benjamin Lu, Johanna Hardin
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace boosting bias correction with the out-of-bag weighted bias correction | random forest · Servo | MSPE: 11.779 -> 17.601 | Hurt |
| apply weighted OOB bias correction | random forest · Step (synthetic) | MSB: 0.814 -> 0.222 | Improved |
| apply weighted OOB bias correction | random forest · Friedman (synthetic) | MSPE: 7.018 -> 4.927 | Improved |
| apply weighted OOB bias correction | random forest · Boston | MSPE: 8.001 -> 6.973 | Improved |
| apply weighted OOB bias correction | random forest · Baseline (synthetic) | MSPE: 1.063 -> 1.085 | No clear effect |
| increase response variance to 100 (noisy setting) | bias-corrected RF vs boosting · Friedman (noised) | MSPE: 119.532 -> 113.898 | Improved |
| weight OOB errors by cohabitation instead of equal weights | prediction intervals · Clustered | interval width: 15.83 -> 13.94 | Improved |
| use OOB error quantiles instead of response quantiles (QRF) | prediction intervals · Friedman | interval width: 36.27 -> 22.01 | Improved |
| use stringent sample-split estimator with three times the training data (ablation) | stringent bias-corrected RF · Friedman (synthetic) | MSPE: 7.327 -> 4.446 | Improved |
| use stringent sample-split estimator with three times the training data (ablation) | stringent PI estimator · Clustered | interval width: 16.57 -> 13.87 | Improved |
| adapt the method to generalized random forest trees | PI estimator · Clustered | interval width: 41.27 -> 11.94 | Improved |
| adapt the method to generalized random forest trees | PI estimator · Boston | coverage: 0.994 -> 0.965 | Improved |
Closest papers by meaning
- Confidence and Uncertainty Assessment for Distributional Random Forests (JMLR 2023 · similarity 0.89)
- Quantifying Uncertainty in Random Forests via Confidence Intervals and Hypothesis Tests (JMLR 2016 · similarity 0.88)
- Stability of Random Forests and Coverage of Random-Forest Prediction Intervals (NeurIPS 2023 · similarity 0.88)
- Generalized Random Forests Using Fixed-Point Trees (ICML 2025 · similarity 0.86)
- DynFrs: An Efficient Framework for Machine Unlearning in Random Forest (ICLR 2025 · similarity 0.84)
- Partition Tree: Conditional Density Estimation over General Outcome Spaces (NeurIPS 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
