This archive contains two files: USArrests-mgcv.R and USArrests-corrected.csv...
This archive contains all R-code needed to invoke functions within current CRAN packages to make all calculations and display all Figures in my 2022 paper Efficient Generalized Ridge Regression published in Open Statistics, 3, 1-18.
The RXshrink R-package for maximum likelihood shrinkage via "generalized ridge" or "least angle" regression methods is currently Version 2.1, March-April 2022. While visiting CRAN, Read about this new version and View / Download the 37-page RXshrink-manual.pdf that documents the many individual functions and datasets within this package.
This REVISED 51 page document contains a wealth of information about GRR estimation with Version 2.2 of RXshrink (Sept. 2022), including examples of new plot function calls, the resulting outputs and details on their interpretation.
This UPDATED 11-page "mcal.pdf" explains why the "multicollinearity allowance" measure of Shrinkage Extent is the IDEAL horizontal axis when plotting ridge TRACE diagnostics.
This revised and updated paper (v3) introduces the Efficient p-parameter Shrinkage Path that heads directly towards and passes directly through the GRR point-estimate that is Most Likely under Normal distribution-theory to achieve an Optimal Variance-Bias Trade-Off in estimation of linear regression coefficients.
ridge, BLUP, Bayes, spline & Stein
1985 - 2004, 185 pages [1781KB]