This repository provides the implementations of the methods described in Tuning-free ridge estimators for high-dimensional generalized linear models.
An R package for the T-Ridge can be found here https://github.com/mohan-zhao/myTridge. Most of the functions are implemented in C++.
We provide an example code in SimulationStudy.Rmd for a comparison of averaged relative prediction errors with 10-fold cross-validated ridge for three types of generalized linear models including Gaussian, Poisson, and Bernoulli cases. Developed for R 3.6.1.
-
Shih-Ting Huang, Ph.D. student in Mathematical Statistics, Ruhr-University Bochum
-
Fang Xie, post-doctoral researcher in Mathematical Statistics, Ruhr-University Bochum
-
Johannes Lederer, Professor in Mathematical Statistics, Ruhr-University Bochum
Additional functions : The source codes of some functions loaded in SimulationStudy.Rmd that are required in the simulation study.
SimulationProcess : The source codes loaded in SimulationStudy.Rmd for generating the simulation results.
RealData : Two data sets, mass-spec500peaks.csv, and riboflavin.csv for applying the t-ridge pipeline on real data and the R codes, RealData.R for the real data analysis.
All of the codes in this repository are written in R and supports all plarforms which are supported by R itself.
This repository depends on R libraries glmnet, MASS, htmlTable, and pander.
The HDIM package is licensed under the MIT license. To
view the MIT license please consult LICENSE.txt.
Tuning-free ridge estimators for high-dimensional generalized linear models
Cite as "S. Huang, F. Xie, and J. Lederer. Tuning-free ridge estimators for high-dimensional generalized linear models. arXiv:2002.11916".