Estimate conditional average treatment effect

Performs CCIT algorithm to estimate conditional average treatment effect.

CCIT(
  matched.exploration.sample.outcomes,
  matched.validation.sample.outcomes,
  matched.inference.sample.outcomes,
  lambdas,
  stopping.rule
)

Arguments

matched.exploration.sample.outcomes

The exploration dataframe with outcome Y, exposure variable treat.

matched.validation.sample.outcomes

The validation dataframe with outcome Y, exposure variable treat.

matched.inference.sample.outcomes

The inference dataframe with outcome Y, exposure variable treat.

lambdas

A vector of values to use as the regularization parameter in the CCIT algorithm.

stopping.rule

A boolean value to indicate whether the tree-splitting algorithm should stop when the estimated interaction effect is <1/10 of the overall effect.

Value

list of: - est.treatment.effects: For each lambda value, dataframe with CATE estimates for all observations - selected.trees: For each lambda value, selected decision tree - tree.list: all decisions trees in the sequence - selected.tree.size: For each lambda value, size of selected decision tree