Plot Rank Average Treatment Effect — margot_plot

This function creates a ggplot visualisation of the Rank Average Treatment Effect. It displays the estimate with a confidence interval, using a simple black line and light gray shading by default. Can either plot an existing RATE object or compute RATE on-demand from a causal forest.

Usage

margot_plot_rate(
  x,
  outcome_var,
  title = NULL,
  subtitle = "(95% confidence interval shown as shaded area)",
  x_lab = "Treated fraction (q)",
  y_lab = "Estimate",
  remove_tx_prefix = TRUE,
  remove_z_suffix = TRUE,
  use_title_case = TRUE,
  remove_underscores = TRUE,
  label_mapping = NULL,
  target = "AUTOC",
  tau_hat = NULL,
  q = seq(0.1, 1, by = 0.1),
  policy = "treat_best",
  subset = NULL,
  use_oob_predictions = TRUE,
  seed = 12345,
  ...
)

Arguments

x: Either an object of class "rank_average_treatment_effect" or a causal_forest object from grf. If a causal_forest is provided, RATE will be computed on-demand.
outcome_var: A character string specifying the name of the outcome variable to plot. This is used for the plot title with proper label transformation.
title: Character string for the plot title. If NULL (default), a title is generated using the transformed outcome variable name.
subtitle: Character string for the plot subtitle. Default explains the confidence interval.
x_lab: Character string for the x-axis label. Default is "Treated fraction (q)".
y_lab: Character string for the y-axis label. Default is "Estimate".
remove_tx_prefix: Logical value indicating whether to remove the "tx_" prefix from labels. Default is TRUE.
remove_z_suffix: Logical value indicating whether to remove the "_z" suffix from labels. Default is TRUE.
use_title_case: Logical value indicating whether to convert labels to title case. Default is TRUE.
remove_underscores: Logical value indicating whether to remove underscores from labels. Default is TRUE.
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.
target: Character; either "AUTOC" or "QINI". Only used when x is a causal_forest. Default is "AUTOC".
tau_hat: Optional vector of treatment effect estimates. Only used when x is a causal_forest. If NULL, will be computed using out-of-bag predictions.
q: Numeric vector of quantiles at which to evaluate. Default is seq(0.1, 1, by = 0.1). Only used when x is a causal_forest.
policy: Character; either "treat_best" or "withhold_best". Only used when x is a causal_forest. Default is "treat_best".
subset: Optional indices for subsetting the evaluation data. Only used when x is a causal_forest.
use_oob_predictions: Logical; if TRUE and tau_hat is NULL, use out-of-bag predictions. Only used when x is a causal_forest. Default is TRUE.
seed: Optional random seed for reproducibility. Only used when x is a causal_forest.
...: Additional arguments passed to ggplot.

Value

A ggplot object that can be further customised or printed.

Examples

if (FALSE) { # \dontrun{
# Assuming rate_eval is your rank_average_treatment_effect object
p <- margot_plot_rate(rate_eval, "model_t2_belong_z")
print(p)

# Using custom label mapping
label_mapping <- list(
  "t2_env_not_env_efficacy_z" = "Deny Personal Environmental Efficacy",
  "t2_env_not_climate_chg_real_z" = "Deny Climate Change Real"
)
p <- margot_plot_rate(rate_eval, "model_t2_env_not_env_efficacy_z",
  label_mapping = label_mapping
)
print(p)
} # }