Skip to contents

Recalculate Policy Trees with Custom Covariates

Source: R/margot_recalculate_policy_trees.R
margot_recalculate_policy_trees.Rd

This function allows recalculation of policy trees using custom covariate sets. It provides flexibility to explore different covariate specifications for policy tree estimation while maintaining the original causal forest results.

Usage

margot_recalculate_policy_trees(
  model_results,
  outcomes_to_recalculate = NULL,
  custom_covariates = NULL,
  exclude_covariates = NULL,
  covariate_mode = c("original", "custom", "add", "all"),
  model_prefix = "model_",
  verbose = TRUE,
  parallel = FALSE,
  n_cores = future::availableCores() - 1,
  seed = 12345,
  tree_method = c("policytree", "fastpolicytree")
)

Arguments

model_results

A list containing the model results from margot_causal_forest().

outcomes_to_recalculate

Character vector of outcome names to recalculate. If NULL (default), recalculates all models marked as needing recalculation.

custom_covariates

Character vector of covariate names to use for policy trees. If NULL, uses the original top variables from the model.

exclude_covariates

Character vector of covariate names or patterns to exclude. Supports exact matches and regex patterns (e.g., "^t0_log_" excludes all variables starting with "t0_log_").

covariate_mode

Character string specifying how to handle covariates:

  • "original"Use original top variables from model (default)

  • "custom"Use only the specified custom_covariates

  • "add"Add custom_covariates to existing top variables

  • "all"Use all available covariates

model_prefix

Character string prefix for model names. Default is "model_".

verbose

Logical indicating whether to display messages. Default is TRUE.

parallel

Logical indicating whether to use parallel processing. Default is FALSE.

n_cores

Number of cores for parallel processing. Default is availableCores() - 1.

seed

Random seed for reproducibility. Default is NULL.

tree_method

Character string specifying which method to use for policy tree computation:

  • "policytree"Use the policytree package (default)

  • "fastpolicytree"Use the fastpolicytree package (about 10x faster)

Value

A modified copy of model_results with recalculated policy trees.

Details

This function is useful for:

  • Exploring sensitivity of policy recommendations to covariate selection

  • Excluding covariates with measurement error or other concerns

  • Using domain knowledge to select policy-relevant covariates

  • Testing robustness of policy tree findings

The function preserves all original model results and only updates:

  • policy_tree_depth_1: Single-split policy tree

  • policy_tree_depth_2: Two-split policy tree (if ≥2 covariates)

  • plot_data: Data for visualization

  • Metadata about which covariates were used

Examples

if (FALSE) { # \dontrun{
# recalculate with custom covariates
results_custom <- margot_recalculate_policy_trees(
  model_results,
  custom_covariates = c("age", "gender", "income"),
  covariate_mode = "custom"
)

# exclude measurement error variables
results_clean <- margot_recalculate_policy_trees(
  model_results,
  exclude_covariates = c("bmi", "^log_"),
  covariate_mode = "original"
)

# use all covariates
results_all <- margot_recalculate_policy_trees(
  model_results,
  covariate_mode = "all"
)
} # }