Compute Qini Curves for Multi-Arm Treatments — compute_qini_curves_multi

This function computes Qini curves for multi-arm treatment effects using the maq package. It handles various edge cases and provides detailed information about the computation process.

Usage

compute_qini_curves_multi_arm(
  tau_hat,
  Y,
  W_multi,
  W.hat = NULL,
  cost = NULL,
  verbose = TRUE
)

Arguments

tau_hat: A matrix or 3D array of estimated treatment effects. For multi-arm treatments, this should be a matrix where each column represents a treatment arm.
Y: A vector or matrix of observed outcomes.
W_multi: A factor vector of treatment assignments for multi-arm treatments.
W.hat: Optional vector specifying known treatment assignment probabilities for each arm.
cost: Optional vector specifying the cost associated with each treatment arm.
verbose: Logical indicating whether to display detailed messages during execution. Default is TRUE.

Value

A list containing:

qini_data: A data frame containing Qini curve data for plotting.
qini_objects: A list of maq objects for each curve, which can be used to compute average gain.

The qini_data has an attribute "imputed" which is TRUE if any curves were imputed with zeros.

Details

The function computes Qini curves for all arms combined, a baseline (no covariates), and each individual treatment arm. It handles cases where some or all Qini objects have zero length or are NULL, extending curves with zeros when necessary.