Description:

This package contains functions to estimate linear and nonlinear regression methods with Penalized Partial Least Squares. Partial Leasts Squares (PLS) is a regression method that constructs latent components Xw from the data X with maximal covariance to a response y. The components are then used in a least-squares fit instead of X. For a quadratic penalty term on w, Penalized Partial Least Squares constructs latent components that maximize the penalized covariance. Applications include the estimation of generalized additive models and functional data.

Features of the package include

• estimation of linear regression models with penalized PLS,
• estimation of generalized additive models with penalized PLS based on splines transformations,
• model selection for both methods based on cross validation.

The package also contains a data set from Near-Infrared Spectroscopy.

For more information on penalized PLS:

N. Krämer, 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.

Please send an email to nkraemer at cs dot tu-berlin dot de for any comments, suggestions, or reports on bugs.

If you want to cite this package, please use the following bib-tex entry.

@Manual{KraBou09,

title = {ppls: Penalized Partial Least Squares},

author = {Nicole Kraemer and Anne-Laure Boulesteix},

year = {2009},

note = {R package version 1.03},

}

Changes to previous version:

• fixed several bugs
• drastic speed-up of computation time

BibTeX Entry: Download

Corresponding Paper BibTeX Entry: Download

Supported Operating Systems: Linux, Macosx, Windows

Data Formats: None

Tags: Kernel, Regression, Partial Least Squares

Archive: download here

Other available revisons

Version	Changelog	Date
1.03	fixed several bugs drastic speed-up of computation time	May 5, 2009, 19:53:20
1.0	Initial Announcement on mloss.org.	October 12, 2007, 15:18:17

Comments

No one has posted any comments yet. Perhaps you'd like to be the first?

Leave a comment

You must be logged in to post comments.