Research map / Probabilistic and scientific modeling / Causal discovery
Decorrelated Variable Importance
Isabella Verdinelli, Larry Wasserman
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 decorrelated parameter psi_3 instead of LOCO psi_L | additive-model estimator · Synthetic Example 2 | confidence interval coverage: 0.00 -> 0.97 | Improved |
| use psi_3 instead of psi_L | additive-model estimator · Synthetic Example 3 | confidence interval coverage: 0.00 -> 0.92 | Improved |
| use psi_3 instead of psi_L | random forest estimator · Synthetic Example 5 | confidence interval coverage: 0.00 -> 1.00 | Improved |
| use psi_3 instead of psi_L | linear-model estimator · Synthetic Example 5 | confidence interval coverage: 0.00 -> 0.00 | No clear effect |
| use psi_3 instead of psi_L | random forest estimator · Synthetic Example 2 | confidence interval coverage: 0.00 -> 0.30 | Hurt |
| use psi_3 instead of psi_L | linear-model estimator · Synthetic Example 3 | confidence interval coverage: 0 -> 0.99 | Improved |
| use psi_0 (nonparametric decorrelated) instead of psi_3 | additive-model estimator · Synthetic Example 4 | confidence interval coverage: 0.20 -> 0.98 | Hurt |
| use psi_0 instead of psi_3 | random forest estimator · Synthetic Example 3 | confidence interval coverage: 0 -> 0.91 | Hurt |
| use psi_2 instead of psi_3 | random forest estimator · Synthetic Example 2 | confidence interval coverage: 0.99 -> 0.30 | Improved |
| use psi_1 instead of psi_3 | additive-model estimator · Synthetic Example 3 | confidence interval coverage: 0.00 -> 0.92 | Hurt |
| use additive instead of linear nuisance models for psi_3 | psi_3 · Synthetic Example 5 | confidence interval coverage: 0.00 -> 0.85 | Improved |
| increase correlation between X and Z1 (delta) | psi_L · Synthetic Example 1 | confidence interval coverage | Hurt |
Closest papers by meaning
- Variable Importance in High-Dimensional Settings Requires Grouping (AAAI 2024 · similarity 0.75)
- Total Variation Floodgate for Variable Importance Inference in Classification (ICML 2024 · similarity 0.74)
- Testing Causal Models with Hidden Variables in Polynomial Delay via Conditional Independencies (AAAI 2025 · similarity 0.74)
- Out-of-Variable Generalisation for Discriminative Models (ICLR 2024 · similarity 0.74)
- Accurate Estimation of Feature Importance Faithfulness for Tree Models (AAAI 2025 · similarity 0.73)
- Feature Attribution with Necessity and Sufficiency via Dual-stage Perturbation Test for Causal Explanation (ICML 2024 · similarity 0.73)
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.
