qtlmt (version 0.1-3)

varGroup: Group variables

Description

Group variables via variable selection such that the grouped variables are optimal for multivariate analysis.

Usage

varGroup(x, z, zscope=NULL, k=qf(0.95,1,nrow(x)-2), kf=k/2,
   method=c("pool","best"), direction=c("both","forward","backward"))

Arguments

x

a data matrix/frame. Columns are variables to select from.

z

a data matrix/frame. Columns are variables with elements 1 or not 1 (any others).

zscope

which variables in z to be used for grouping; all if "NULL".

k

entry/stay value in backward stepwise.

kf

entry/stay value in forward stepwise.

method

grouping method at each step: pool all the groups selected from each zcope variable, or choose the largest group (see examples).

direction

forward selection, backward elimination or both stepwise.

Value

• indicators of grouped variables.

Examples

data(etrait)
varGroup(traits, mdat, zscope=c(42,55), method="pool")
varGroup(traits, mdat, zscope=c(42,55), method="best")