Skip to contents

Adapter around GenGC's doubly robust estimator. Mirrors est_gengc but uses GenGC::gengc_dr with cross-fitting while respecting the airlock and single-thread contract.

Usage

est_gengc_dr(df, tau = cs_tau_oracle, config = list())

Arguments

df

Data frame with y, w, and covariates; oracle columns (y0, y1, p, structural_te) are dropped before calling GenGC.

tau

Numeric vector of quantile levels passed to tau_grid.

config

Optional list. Common fields include:

  • ci_method: CI intent; one of "none", "default", "bootstrap", "native" (see cs_ci_methods).

  • seed: required when bootstrap CIs are requested.

  • n_boot: number of bootstrap draws (default: 200).

  • n_draws (default 300), num_trees (default 800), crossfit_folds (default 5), num_threads (default 1, enforced).

  • method: GenGC engine; one of "qrf" (default) or "qr".

  • screen: logical; whether to enable GenGC screening (default: FALSE).

For this estimator, ci_method = "default" maps to "bootstrap".

Value

A legacy-compatible estimator result containing scalar att, a QST table at the requested tau values, cf = NULL, and CI/provenance meta.