Penalized PLS based on NIPALS Algorithm and blockswise variable selection

Description

Internal function that computes the penalized PLS solutions with included block-wise variable selection.

Usage

1
penalized.pls.select(X, y, M, ncomp,blocks)

Arguments

X

matrix of centered and (possibly) scaled input data

y

vector of centered and (possibly) scaled response data

M

matrix that is a transformation of the penalty term P. Default is M=NULL, which corresponds to no penalization.

ncomp

number of PLS components

blocks

vector of length ncol(X) that encodes the block structure of X.

Details

This function assumes that the columns of X and y are centered and - optionally - scaled. The matrix M is defined as the inverse of (I + P) . The computation of the regression coefficients is based on an extension of the classical NIPALS algorithm for PLS. Moreover, in each iteration, the weight vector is only defined by one block of variables. For more details, see Kraemer, Boulesteix, and Tutz (2008).

Value

coefficients

Penalized PLS coefficients for all 1,2,...,ncomp components

Note

This is an internal function that is called by penalized.pls.

Author(s)

Nicole Kraemer

References

N. Kraemer, A.-L. Boulsteix, and G. Tutz (2008). Penalized Partial Least Squares with Applications to B-Spline Transformations and Functional Data. Chemometrics and Intelligent Laboratory Systems, 94, 60 - 69. http://dx.doi.org/10.1016/j.chemolab.2008.06.009

See Also

penalized.pls, ppls.splines.cv

Examples

1
# this is an internal function