This package got retired and archived on 2023-05-05

You can still find the gfn-function package file here: https://github.com/atecon/ridge/blob/master/src/ridge.gfn

Please use the regls addon written by Allin Cottrell instead:

https://sourceforge.net/projects/gretl/files/addons/doc/regls.pdf

Introduction

Package for running Ridge-regression using gretl.

Authors: Sven Schreiber and Artur Tarassow

Description

This does not involve any cross-validation for obtaining some optimal shrinkage hyper-parameter 'lambda'.

Point coefficients as well as associated standard errors are returned for pre-defined lambda values.

Various statistics-of-fit are computed:

1. R-square as computed in scikit-learn
2. R-square as the squared correlation between reaizations and fitted values
3. Adjusted version of (2)
4. AIC
5. BIC
6. Generalized Cross-Validation statistics