select.varimportance: Automatic selection of variables

In chavent/modvarsel: Implements a computational methodology for model and variables selection

Description

Chooses automaticaly the covariates relevant to predict the response variable. By default a change point criterion is used to define a cutoff value and the covariates with mean VI above this cutoff value are selected. It is also possible to choose the number of covariates to select rather than using the changepoint cutoff value.

Usage

## S3 method for class 'varimportance'
select(imp, cutoff = TRUE, nbsel = NULL, ...)

Arguments

imp	an object of class varimportance.
cutoff	if TRUE covariates above the changepoint cutoff value are selected.
nbsel	the number of covariates to select (only if cutoff=FALSE).
...	further arguments passed to or from other methods.

Value

an object of class list with two components.

var	a vector with the names of the selected variables
indices	a vector with the indices of the selected variables

See Also

varimportance, plot.varimportance

Examples

data("simus")
X <- simus$X
Y <- simus$Y1
imp <- varimportance(X,Y,method="linreg",nperm=8)
select(imp, cutoff=TRUE) # by default
select(imp, cutoff=FALSE, nbsel=7)

chavent/modvarsel documentation built on May 22, 2019, 2:22 p.m.