estimate: Causal effect estimation via plug-in g-computation.

In funcml: Functional Machine Learning Framework

Causal effect estimation via plug-in g-computation.

Description

Causal effect estimation via plug-in g-computation.

Usage

estimate(
  data,
  formula,
  model = NULL,
  treatment = NULL,
  estimand = c("ATE", "ATT", "CATE", "IATE"),
  newdata = NULL,
  treatment_level = NULL,
  control_level = NULL,
  spec = NULL,
  type = NULL,
  interval = c("normal", "bootstrap"),
  conf_level = 0.95,
  n_boot = 200,
  seed = NULL,
  fit = NULL,
  ...
)

Arguments

data	Data frame.
formula	Outcome model formula. The first term on the right-hand side is treated as the treatment variable unless treatment is supplied.
model	Learner id (ignored if fit supplied).
treatment	Optional treatment variable name.
estimand	One of "ATE", "ATT", "CATE", or "IATE".
newdata	Optional target population for estimand = "CATE" or "IATE".
treatment_level	Optional treated level for binary treatment.
control_level	Optional control level for binary treatment.
spec	Hyperparameter list passed to fit().
type	Prediction type override for the outcome model.
interval	Interval method: "normal" or "bootstrap".
conf_level	Confidence level for uncertainty intervals.
n_boot	Number of bootstrap resamples used when interval = "bootstrap".
seed	Optional seed.
fit	Optional preconfigured funcml_fit object.
...	Passed to fit().

Value

A funcml_estimand object.

Examples

causal_data <- mtcars
causal_data$am <- factor(causal_data$am, labels = c("auto", "manual"))
ate <- estimate(
  data = causal_data,
  formula = mpg ~ am + wt + hp,
  model = "glm",
  treatment = "am",
  estimand = "ATE"
)
ate$estimate

funcml documentation built on Aug. 22, 2026, 5:08 p.m.