qtlmt (version 0.1-1)

varStep: Variable selection

Description

Add a variable, drop a variable, or select a subset of variables via variable selection that are optimal for multivariate analysis.

Usage

varAdd1(x, grp, vin=NULL, scope=1:ncol(x), k=0)

varDrop1(x, grp, vin=1:ncol(x), k=0)

varStep(x, grp, scope, k, kf=k/2, direction=c("both",
   "forward","backward"))

Arguments

x

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

grp

grouping indicator of observations.

vin

which variables (i.e., columns of x) already in model. It defines the initial model.

scope

which variables (i.e., columns of x) to select from.

k

entry/stay value in backward stepwise.

kf

entry/stay value in forward stepwise.

direction

forward selection, backward elimination or both stepwise.

Value

• which variable to add (add1), which variable to drop (drop1), or a subset of variables in the final model (step).

See Also

varGroup and varSelect

Examples

data(etrait)
varAdd1(traits, grp=mdat[,42], vin=10, scope=1:ncol(traits))
varStep(traits, grp=mdat[,42], k=12, scope=1:ncol(traits),
   direction="back")