Skip to contents

Summarize a fitted Drmatch object on new data

summary.Drmatch.Rd

Generates predicted treatment decisions from a fitted Drmatch object and summarizes their performance against observed treatment assignments in a new dataset. Depending on the requested stage, the function evaluates stage-1 decisions, stage-2 decisions, or both jointly.

Usage

# S3 method for class 'Drmatch'
summary(
  object,
  newdata,
  stage = c("both", "stage1", "stage2"),
  overall_type = c("macro", "weighted"),
  obs1_var = NULL,
  obs2_var = NULL,
  eta2_var = NULL,
  pred1_var = "A1.opt",
  pred2_var = "A2.opt",
  Tc1_var = NULL,
  Tt1_var = NULL,
  Tc2_var = NULL,
  Tt2_var = NULL,
  Tc_total_var = NULL,
  Tt_total_var = NULL,
  as.data.frame = TRUE,
  ...
)

Arguments

object

A fitted object of class "Drmatch".

newdata

A data frame on which to compute predicted treatment assignments and performance summaries.

stage

Character string indicating which decision stage to summarize. One of "both", "stage1", or "stage2". Default is "both".

overall_type

Character string indicating the type of overall summary metrics to compute for the joint path. One of "macro" or "weighted". Default is "macro".

obs1_var

Optional character string giving the observed stage-1 treatment column in newdata. If NULL, the value is taken from object$A1.var.

obs2_var

Optional character string giving the observed stage-2 treatment column in newdata. If NULL, the value is taken from object$A2.var.

eta2_var

Optional character string giving the stage-2 eligibility column in newdata. If NULL, the value is taken from object$eta2.var.

pred1_var

Character string giving the column name in the prediction output corresponding to the estimated stage-1 treatment. Default is "A1.opt".

pred2_var

Character string giving the column name in the prediction output corresponding to the estimated stage-2 treatment. Default is "A2.opt".

Tc1_var

Optional character string giving the stage-1 value column for treatment option -1.

Tt1_var

Optional character string giving the stage-1 value column for treatment option 1.

Tc2_var

Optional character string giving the stage-2 value column for treatment option -1.

Tt2_var

Optional character string giving the stage-2 value column for treatment option 1.

Tc_total_var

Optional character string giving the total value column for treatment option -1.

Tt_total_var

Optional character string giving the total value column for treatment option 1.

as.data.frame

Logical; if TRUE, the output is returned as a one-row data frame. Otherwise, a named numeric vector is returned. Default is TRUE.

...

Additional arguments passed to predict.Drmatch().

Value

Either a one-row data frame or a named numeric vector containing the policy performance metrics returned by policy_summary_metrics().

Details

Internally, predicted treatment assignments are passed to policy_summary_metrics() to compute stage-specific and joint-path classification metrics, with optional value-based summaries when outcome columns are supplied.

When stage = "stage1", stage-2 predictions are set to NA before summary metrics are computed. When stage = "stage2", the observed stage-1 treatment from newdata is used as the stage-1 component of the joint path, while stage-2 predictions are taken from predict(object, ...).

This method requires newdata because performance is evaluated by comparing predicted treatment decisions with observed treatments and, optionally, with value columns provided in the new dataset.

See also

policy_summary_metrics(), print.summary.Drmatch(), predict.Drmatch()

Examples

if (FALSE) { # \dontrun{
# fit <- Drmatch(...)
# summary(fit, newdata = test_dat)
# summary(fit, newdata = test_dat, stage = "stage1")
# summary(fit, newdata = test_dat, stage = "stage2", as.data.frame = FALSE)
} # }