Skip to contents

Reference

Source: vignettes/reference.Rmd
reference.Rmd

API Summary

Function Purpose
fit_observational_forest() Heterogeneous treatment effect from observational contrast
fit_survival_forest() Heterogeneous effect for right-censored outcomes
fit_instrumental_forest() Heterogeneous local effect surrogate in IV setting
rank_effects() Return top or bottom estimated effects
as.data.frame() Extract effect table
plot() Quick histogram of estimated effects

Returned Object

heteff_fit contains:

  • type
  • fit (raw grf forest object)
  • effect_table (sample, estimate, std.error)
  • variable_importance
  • columns (input mapping)

Internal Process Names

The package keeps an intentionally short internal process:

  1. validate input columns
  2. cast covariates to numeric matrix
  3. call the selected grf forest
  4. build a common effect table
  5. return unified heteff_fit object

Typical Integration Pattern 

fit <- fit_observational_forest(df, "outcome", "treatment", c("x1", "x2"))
tab <- as.data.frame(fit)
top <- rank_effects(fit, n = 25)