Batch Processing of Policy Trees and Related Visualizations

Usage

margot_policy(
  result_outcomes,
  policy_tree_args = list(),
  decision_tree_args = list(),
  dpi = 600,
  width = 12,
  height = 12,
  save_plots = TRUE,
  output_dir = here::here(push_mods),
  spend = c(0.2, 0.5),
  label_mapping = NULL,
  original_df = NULL
)

Arguments

result_outcomes: A list containing the results from margot_multi_arm_causal_forest().
policy_tree_args: A list of arguments to pass to margot_plot_policy_tree(). Default is list().
decision_tree_args: A list of arguments to pass to margot_plot_decision_tree(). Default is list().
dpi: The resolution of saved plots in dots per inch. Default is 600.
width: The width of saved plots in inches. Default is 12.
height: The height of saved plots in inches. Default is 12.
save_plots: Logical indicating whether to save plots to disk. Default is TRUE.
output_dir: The directory to save plots in. Default is here::here(push_mods).
spend: A vector of spend levels to use for difference gain summaries. Default is c(0.2, 0.5).
label_mapping: Optional named list for custom label mappings. Keys should be original variable names (with or without "model_" prefix), and values should be the desired display labels. Default is NULL.
original_df: Optional dataframe with untransformed variables, used to display split values on the data scale.

Value

A list where each element corresponds to a model in the input `result_outcomes`. Each element is itself a list containing:

policy_tree_plot: A ggplot object of the policy tree plot
policy_tree_interpretation: A character string interpreting the policy tree
qini_plot: A ggplot object of the Qini plot
decision_tree_visualisation: A ggplot object visualizing the decision tree
policy_combo_plot: A ggplot object of the policy combo plot
diff_gain_summaries: A nested list containing difference gain summaries for each spend level