rGEDI (version 0.6.0)

varSel: Select predictor variables for a GEDI model

Description

Select predictor variables for a GEDI model

Usage

varSel(
  x,
  y,
  method = c("correlation", "randomForest", "rfe"),
  threshold = 0.1,
  seed = NULL,
  ...
)

Value

An object of class gedi_var_selection.

Arguments

x

Predictor data frame.

y

Response vector.

method

Selection method. "rfe" recursively removes the least important predictor and selects the smallest model whose out-of-bag error is within one standard error of the minimum.

threshold

Minimum absolute correlation or scaled Random Forest importance. For method = "rfe", it is applied after model selection; use 0 to retain the RFE-selected subset.

seed

Optional random seed.

...

Passed to randomForest when requested.