predict.causalBoosting: Make predictions from a fitted causal boosting model

In saberpowers/causalLearning: Methods for heterogeneous treatment effect estimation

Description

Make predictions from a fitted causal boosting model

Usage

## S3 method for class 'causalBoosting'
predict(object, newx, newtx = NULL,
  type = c("treatment.effect", "conditional.mean", "response"),
  num.trees = 1:object$num.trees, honest = FALSE, naVal = 0, ...)

Arguments

object	a fitted causalBoosting object
newx	matrix of new covariates for which estimated treatment effects are desired
newtx	option vector of new treatment assignments (only used if type = 'response')
type	type of prediction required: 'treatment.effect' returns estimated treatment effect. 'conditional.mean' returns two predictions, one for each arm. 'response' returns prediction for arm corresponding to newtx.
num.trees	number(s) of shallow causal trees to use for prediction
honest	logical: should honest re-estimates of leaf means be used for prediction? This requires that x.est, tx.est, y.est were specified when the causal boosting model was fit
naVal	value with which to replace NA predictions
...	ignored

Value

a vector or matrix of predictions corresponding to the rows of newx

saberpowers/causalLearning documentation built on May 30, 2019, 8:26 a.m.