Doubly-robust ATE via causal forest (grf) — morie_estimate_dr

Native rmorie causal forest: the R-learner decomposition (Nie & Wager 2021) with cross-fit nuisances and a weighted subsampled regression forest for tau(x), combined through the AIPW orthogonal score — the same estimand grf's average_treatment_effect(method = "AIPW") targets (cross-validated against grf in the package's cross tests). A machine-learning alternative to the GLM-based morie_estimate_aipw(). No grf at runtime.

Usage

morie_estimate_dr_forest(
  data,
  treatment,
  outcome,
  covariates,
  target_sample = c("all", "treated", "control", "overlap")
)

Arguments

data: A data frame.
treatment: Name of the binary treatment column.
outcome: Name of the outcome column.
covariates: Character vector of covariate names.
target_sample: Sample to average over, passed to grf: one of "all", "treated", "control", "overlap".

Value

A list with ate, se, ci_lower, ci_upper, n.

Examples

set.seed(1)
df <- data.frame(t = rbinom(200, 1, 0.4), y = rnorm(200), x = rnorm(200))
morie_estimate_dr_forest(df, "t", "y", "x")
#> $ate
#> [1] -0.04573452
#> 
#> $se
#> [1] 0.1424301
#> 
#> $ci_lower
#> [1] -0.3248976
#> 
#> $ci_upper
#> [1] 0.2334285
#> 
#> $n
#> [1] 200
#>