grf (version 1.2.0)

tune_ll_causal_forest: Local linear forest tuning

Description

Finds the optimal ridge penalty for local linear causal prediction.

Usage

tune_ll_causal_forest(
  forest,
  linear.correction.variables = NULL,
  ll.weight.penalty = FALSE,
  num.threads = NULL,
  lambda.path = NULL
)

Arguments

forest

The forest used for prediction.

linear.correction.variables

Variables to use for local linear prediction. If left null, all variables are used. Default is NULL.

ll.weight.penalty

Option to standardize ridge penalty by covariance (TRUE), or penalize all covariates equally (FALSE). Defaults to FALSE.

num.threads

Number of threads used in training. If set to NULL, the software automatically selects an appropriate amount.

lambda.path

Optional list of lambdas to use for cross-validation.

Value

A list of lambdas tried, corresponding errors, and optimal ridge penalty lambda.

Examples

# NOT RUN {
# Find the optimal tuning parameters.
n <- 50
p <- 10
X <- matrix(rnorm(n * p), n, p)
W <- rbinom(n, 1, 0.5)
Y <- pmax(X[, 1], 0) * W + X[, 2] + pmin(X[, 3], 0) + rnorm(n)

forest <- causal_forest(X, Y, W)
tuned.lambda <- tune_ll_causal_forest(forest)

# Use this parameter to predict from a local linear causal forest.
predictions <- predict(forest, linear.correction.variables = 1:p,
                       ll.lambda = tuned.lambda$lambda.min)
# }
# NOT RUN {
# }