ForestIV: ForestIV Main Function
In mochenyang/ForestIV: EnsembleIV

ForestIV

R Documentation

ForestIV Main Function

This function implements the main ForestIV approach.

ForestIV(
  data_test,
  data_unlabel,
  control,
  method,
  iterative = TRUE,
  ntree,
  model_unbias,
  family,
  diagnostic,
  select_method
)

`data_test`	Testing dataframe for random forest, must have a column named "actual" that contains the ground truth, and all trees' predictions.
`data_unlabel`	Unlabel dataframe for random forest, must have all trees' predictions.
`control`	A character vector of control variable names. Pass an empty vector if there are no control variables
`method`	"Lasso" for ForestIV method and "IIV" for EnsembleIV method.
`iterative`	Whether to perform iterative IV selection or not, default to TRUE. Only relevant when method = "Lasso"
`ntree`	Number of trees in the random forest.
`model_unbias`	Unbiased estimation.
`family`	Model specification, same as in the family parameter in glm.
`diagnostic`	Whether to output diagnostic correlations for instrument validity and strength, default to TRUE.
`select_method`	method of IV selection. One of "optimal" (LASSO based), "top3", and "PCA".