grf (version 1.2.0)

average_partial_effect: Estimate average partial effects using a causal forest

Description

Gets estimates of the average partial effect, in particular the (conditional) average treatment effect (target.sample = all): 1/n sum_i = 1^n Cov[Wi, Yi | X = Xi] / Var[Wi | X = Xi]. Note that for a binary unconfounded treatment, the average partial effect matches the average treatment effect.

Usage

average_partial_effect(
  forest,
  calibrate.weights = TRUE,
  subset = NULL,
  debiasing.weights = NULL,
  num.trees.for.variance = 500
)

Arguments

forest

The trained forest.

calibrate.weights

Whether to force debiasing weights to match expected moments for 1, W, W.hat, and 1/Var[W|X].

subset

Specifies a subset of the training examples over which we estimate the ATE. WARNING: For valid statistical performance, the subset should be defined only using features Xi, not using the treatment Wi or the outcome Yi.

debiasing.weights

A vector of length n (or the subset length) of debiasing weights. If NULL (default) these are estimated by a variance forest.

num.trees.for.variance

Number of trees used to estimate Var[Wi | Xi = x]. Default is 500. (only applies when debiasing.weights = NULL)

Value

An estimate of the average partial effect, along with standard error.

Details

If clusters are specified, then each unit gets equal weight by default. For example, if there are 10 clusters with 1 unit each and per-cluster ATE = 1, and there are 10 clusters with 19 units each and per-cluster ATE = 0, then the overall ATE is 0.05 (additional sample.weights allow for custom weighting). If equalize.cluster.weights = TRUE each cluster gets equal weight and the overall ATE is 0.5.

Double robust scores are calculated with a component based on a forest estimate of Var[Wi | Xi = x]. These weights can also be passed manually by specifying debiasing.weights.

Examples

# NOT RUN {
n <- 2000
p <- 10
X <- matrix(rnorm(n * p), n, p)
W <- rbinom(n, 1, 1 / (1 + exp(-X[, 2]))) + rnorm(n)
Y <- pmax(X[, 1], 0) * W + X[, 2] + pmin(X[, 3], 0) + rnorm(n)
tau.forest <- causal_forest(X, Y, W)
tau.hat <- predict(tau.forest)
average_partial_effect(tau.forest)
average_partial_effect(tau.forest, subset = X[, 1] > 0)
# }
# NOT RUN {
# }