imptree_control {imptree}

Control parameters for generating imptree objects

Description

Initializing and validating the tree generation parameters

Usage

imptree_control(splitmetric, controlList = NULL, tbase = 1,
  gamma = 1, depth = NULL, minbucket = 1L, ...)

Arguments

splitmetric

Choosen split metric as integer: 0 means "globalmax" and 1L "range", repectively. See imptree_params

controlList

Named list containing the processed arguments. See details.

tbase

Value that needs to be at least attained to qualify for splitting (default: 1)

gamma

Weighting factor of the maximum entropy (default: 1)

depth

Integer limiting the tree to the given depth, with 0 indicating to perform no splitting at all. If not supplied, NULL (default) or negative the tree is grown to maximal size, the latter triggering a warning.

minbucket

Positive integer as minimal leaf size (default: 1)

...

Argument gobbling; is not processed

Details

The argument controlList may be a named list with names in c("tbase", "gamma", "depth", "minbucket") Any values in this list will overwrite those supplied in named arguments. When controlList = NULL (default) only the supplied arguments are checked.

In case controlList contains an argument named splitmetric, this will be ignored. If splitmetric is 0L, i.e. "globalmax", the values for gamma and tbase are set to their default values, even if the user supplied different values.

Value

A list containing the options. Missing options are set to their default value.

Author(s)

Paul Fink [e-mail]

See Also

imptree, imptree_params

Examples

## Check performed for splitmetric 'globalmax',
## tbase' is default generated and 'gamma' is overwritten
## (see Details), tree is grown to full depth and 
## at least 5 observations are needed to be within each node
imptree_control(splitmetric = 0, gamma = 0.5,
                depth = NULL, minbucket = 5)

## Passing some control arguments in a list
## As splitmetric is 'range', gamma is respected
imptree_control(splitmetric = 1, minbucket = 5,
                controlList = list(gamma = 0.5, depth = NULL))