Skip to contents

Teacher models for causal distillation trees

teacher_models.Rd

These functions are wrappers around various heterogeneous treatment effect learners (and their associated predict methods) that can be easily used as teacher models in the causal distillation tree framework.

  • causal_forest(): wrapper around grf::causal_forest()

  • predict_causal_forest(): wrapper around predict() for causal_forest() models.

  • bcf(): wrapper around bcf::bcf()

  • predict_bcf(): wrapper around predict() for bcf() models.

  • rlearner_teacher(): wrapper around model functions from the rlearner package to convert them to teacher models for CDT.

  • rboost(): (defunct) wrapper around rlearner::rboost().

  • rlasso(): (defunct) wrapper around rlearner::rlasso().

  • rkern(): (defunct) wrapper around rlearner::rkern().

Warning: The rboost(), rlasso(), and rkern() functions are defunct as of version 1.0.0. Use rlearner_teacher() (e.g., rlearner_teacher(rlearner::rboost)) instead to convert rlearner functions into correct format for use as teacher model in CDT.

Usage

causal_forest(X, Y, Z, W = NULL, ...)

predict_causal_forest(...)

rlearner_teacher(rlearner_fun, ...)

rboost(X, Y, Z, W = NULL, ...)

rlasso(X, Y, Z, W = NULL, ...)

rkern(X, Y, Z, W = NULL, ...)

bcf(
  X,
  Y,
  Z,
  W = NULL,
  pihat = "default",
  w = NULL,
  nburn = 2000,
  nsim = 1000,
  n_threads = 1,
  ...
)

predict_bcf(...)

Arguments

X

A tibble, data.frame, or matrix of covariates.

Y

A vector of outcomes.

Z

A vector of treatments.

W

A vector of weights corresponding to treatment propensities.

...

Additional arguments to pass to the base model functions.

rlearner_fun

One of rlearner::rboost, rlearner::rlasso, or rlearner::rkern to be transformed to teacher model format for CDT.

pihat

Length n estimates of propensity score

w

An optional vector of weights. When present, BCF fits a model \(y | x ~ N(f(x), \sigma^2 / w)\), where \(f(x)\) is the unknown function.

nburn

Number of burn-in MCMC iterations

nsim

Number of MCMC iterations to save after burn-in. The chain will run for nsim*nthin iterations after burn-in

n_threads

An optional integer of the number of threads to parallelize within chain bcf operations on