ConfoundingSHAP defines a custom Shapley game to attribute confounding strength to individual covariates and uses TabPFN to estimate it scalably without exhaustive refitting.
arXiv preprint arXiv:2002.11631 (2020)
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
verdicts
UNVERDICTED 8roles
background 1polarities
background 1representative citing papers
Differential subgroups identify specific feature combinations where population differences in outcomes are most extreme, found via a new optimization objective and the DiffSub method.
Large-scale study finds that counterfactual metrics on semi-simulated data do not select the same estimators as observable metrics on real data, and benchmark rankings fail to transfer.
Introduces CounterBench benchmark and CoIn iterative reasoning method showing LLMs perform near random on formal counterfactual tasks but improve substantially with guided backtracking.
Introduces the A2A metric to evaluate propensity score matching methods by constructing artificial tasks with known outcomes, reducing ATE errors up to 50% when combined with SMD.
An integrated framework using ensemble hybrid ranking models and monotonicity-based extrapolation enables personalized policy deployment in a constrained two-sided job marketplace, achieving significant target metric improvements with guardrail compliance.
Spark Policy Toolkit supplies semantic contracts plus mapInPandas/mapInArrow inference and executor-side split search so policy learning remains correct and fast on Spark clusters up to tens of millions of rows.
A software framework integrates heterogeneous causal inference, policy learning, mediation, forecasts, variance reduction, and anytime-valid inference into one AI-orchestratable interface for business experimentation.
citing papers explorer
-
Improving Bias Correction Standards by Quantifying its Effects on Treatment Outcomes
Introduces the A2A metric to evaluate propensity score matching methods by constructing artificial tasks with known outcomes, reducing ATE errors up to 50% when combined with SMD.