Research map / Probabilistic and scientific modeling / Causal discovery
A Simple Unified Approach to Testing High-Dimensional Conditional Independences for Categorical and Ordinal Data
Ankur Ankan, Johannes Textor
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 LS residualization with a Hotelling-style test and GLM conditional probability models instead of stratification-based CI testing | High-dimensional categorical and ordinal CI testing · Simulated binary data, calibration | calibration | Improved |
| Use LS residualization with a Hotelling-style test and random-forest conditional probability models instead of stratification-based CI testing | High-dimensional categorical and ordinal CI testing · Simulated binary data, calibration | calibration | Improved |
| Use GLM-based residualization rather than the compared CI tests | Categorical CI testing with nuisance conditioning variables · Simulated categorical data, discrimination | accuracy | Improved |
| Use random-forest-based residualization rather than the compared CI tests | Categorical CI testing with nuisance conditioning variables · Simulated categorical data, discrimination | accuracy | Improved |
| Use GLM-based residualization rather than JT | Ordinal CI testing · Simulated ordinal data with multiple conditioning variables and small samples | accuracy | Improved |
| Use random-forest-based residualization rather than JT | Ordinal CI testing · Simulated ordinal data with multiple conditioning variables and small samples | accuracy | Improved |
| Use GLM or random-forest residualization rather than stratification-based tests | CI testing implied by random DAGs · Dense DAGs, model testing | precision | Improved |
| Use GLM or random-forest residualization rather than stratification-based tests | PC structure learning · Dense simulated DAGs | F1 score | Improved |
| Use GLM or random-forest residualization rather than the strongest reported baselines | PC structure learning · Sparse simulated DAGs | F1 score | No clear effect |
| Use GLM or random-forest residualization rather than MC-MI | PC structure learning · Alarm | F1 score | Hurt |
| Use GLM or random-forest residualization rather than MC-MI | PC structure learning · Insurance | F1 score | No clear effect |
| Use GLM or random-forest residualization rather than stratification-based testing | PC structure learning · Adult income | F1 score | Improved |
Closest papers by meaning
- PAIR-CI: Calibrated Conditional Independence Testing for Causal Discovery with Incomplete Data (NeurIPS 2026 · similarity 0.90)
- Testing Causal Models with Hidden Variables in Polynomial Delay via Conditional Independencies (AAAI 2025 · similarity 0.88)
- A Conditional Independence Test in the Presence of Discretization (ICLR 2025 · similarity 0.88)
- Causal Discovery via Conditional Independence Testing with Proxy Variables (ICML 2024 · similarity 0.88)
- On the Hardness of Conditional Independence Testing In Practice (NeurIPS 2025 · similarity 0.87)
- A Sample Efficient Conditional Independence Test in the Presence of Discretization (ICML 2025 · 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.
