RXshrink: Maximum Likelihood Shrinkage using Generalized Ridge or Least Angle Regression Methods

Functions are provided to calculate and display ridge TRACEs for various shrinkage PATHS. They determine the m-Extent of shrinkage most likely, under Normal-distribution theory, to produce biased estimates of regression (beta) coefficients with minimum MSE Risk. An alternative function, correct.signs(), provides beta coefficient estimates when an ill-conditioned (nearly multicollinear) model yields OLS estimates with ("wrong") numerical signs that disagree with the signs of the observed correlations between the Y-outcome and the selected X-predictor variables.

Version: 1.3.1
Depends: lars
Published: 2020-01-23
Author: Bob Obenchain
Maintainer: Bob Obenchain <wizbob at att.net>
License: GPL-2
URL: https://www.R-project.org , http://localcontrolstatistics.org
NeedsCompilation: no
In views: MachineLearning
CRAN checks: RXshrink results

Downloads:

Reference manual: RXshrink.pdf
Package source: RXshrink_1.3.1.tar.gz
Windows binaries: r-devel: RXshrink_1.3.1.zip, r-devel-gcc8: RXshrink_1.3.1.zip, r-release: RXshrink_1.3.1.zip, r-oldrel: RXshrink_1.3.1.zip
OS X binaries: r-release: RXshrink_1.3.1.tgz, r-oldrel: RXshrink_1.3.1.tgz
Old sources: RXshrink archive

Linking:

Please use the canonical form https://CRAN.R-project.org/package=RXshrink to link to this page.