Bayesian penalized regression with continuous shrinkage prior densities


We have implemented a new MATLAB toolbox for Bayesian regression. The toolbox features Bayesian linear regression with Gaussian or heavy-tailed error models and Bayesian logistic regression with ridge, lasso, horseshoe and horseshoe+ estimators. Two heavy-tailed error models are supported in linear regression: Laplace and Student t distribution errors. The toolbox is very efficient and can be used with high-dimensional data. This toolbox uses the C/C++ PolyaGamma sampler written by Jesse Windle.

Version 1.1 of the MATLAB toolbox can be downloaded from here. Please see “examples_bayesreg.m” for examples on how to use the toolbox, or type “help bayesreg” within MATLAB.

An R version of this toolbox is now available on CRAN. To install the R package, type “install.packages(“bayesreg”)” within R.

To cite this toolbox:

  1. Makalic E. & Schmidt, D. F.
    High-Dimensional Bayesian Regularised Regression with the BayesReg Package
    arXiv:1611.06649 [stat.CO], 2016

Updates:
(23/01): Updated the bayesreg code to version 1.2. This version implements Zellner’s g-prior for linear and logistic regression. The g-prior only works with full rank matrices. The examples in “examples_bayesreg.m” have been updated to include a g-prior example.

(17/01): Updated the bayesreg code to version 1.1. The display code is now in a separate “summary()” function so that it can be used many times on the same data and sampling results. We have updated “examples_bayesreg.m” to include examples of the new “summary()” function.

  1. No comments yet.
(will not be published)