Research map / Probabilistic and scientific modeling / Molecules and proteins
Quantifying Uncertainty in Random Forests via Confidence Intervals and Hypothesis Tests
Lucas Mentch, Giles Hooker
Read the paperCheck your idea against 137,217 papers
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 |
|---|---|---|---|
| use internal instead of external variance estimate for confidence intervals | subbagged ensemble CIs · SLR and MARS simulations | coverage probability: 0.956 -> 0.936 | No clear effect |
| use internal instead of external variance estimate for the feature significance test | subbagged hypothesis test · MARS irrelevant feature X6 | empirical alpha level: 0 -> 0.14 | Hurt |
| increase subsample size k to 200 (from 75) with n=1000 | subbagged ensemble · MARS simulation | normality of predictions: k=75 -> k=200 | Hurt |
| use full bootstrap samples (k = n = 1000) instead of subsamples | ensemble · MARS simulation | normality of predictions: k=1000 | Hurt |
| estimate zeta_1,k with n_MC=250 instead of larger n_MC | external variance estimation · SLR and MARS simulations | variance estimate: n_MC = 250 vs n_MC = 1000 | Hurt |
| choose test points in the interior of the feature space rather than near edges | chi-square feature test · MARS simulation | fit of test statistics to chi-square | Improved |
| replace month with random values (ablation) | subsampled forest · eBird 2010 | test statistic vs critical value 31.41: 4233.10 -> 58.02 | Hurt |
| add the real month feature versus a random month feature (ablation) | subsampled forest · eBird 2010 | test statistic: 2336.14 | Improved |
| replace year with random values (ablation) | subsampled forest · eBird 2004-2010 | test statistic vs critical value 31.41: 94.43 -> 52.51 | Hurt |
| add the real year feature versus a random year feature (ablation) | subsampled forest · eBird 2004-2010 | test statistic: 109.72 | Improved |
Closest papers by meaning
- Scalable and Efficient Hypothesis Testing with Random Forests (JMLR 2022 · similarity 0.89)
- A Unified Framework for Random Forest Prediction Error Estimation (JMLR 2021 · similarity 0.88)
- Randomization Can Reduce Both Bias and Variance: A Case Study in Random Forests (JMLR 2025 · similarity 0.88)
- Smooth And Consistent Probabilistic Regression Trees (NeurIPS 2020 · similarity 0.88)
- Randomization as Regularization: A Degrees of Freedom Explanation for Random Forest Success (JMLR 2020 · similarity 0.88)
- Smaller, more accurate regression forests using tree alternating optimization (ICML 2020 · similarity 0.87)
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.
