shrinkage

An R package to fit Bayesian regression models with shrinkage priors.

Linear regression models with global shrinkage priors:

• normal-inverse-Gamma
• normal-beta prime (which includes normal–half Cauchy)
• normal-inverse-Gaussian
• normal-Gamma
• normal with empirical Bayes (closed-form inference)

Linear regression models with local shrinkage priors, which accommodate grouped variables and some sparse-inducing priors:

• normal-inverse-Gamma (Armagan and Zaretzki, 2010)
• normal-Gamma (Griffin and Brown, 2010)
• normal-inverse-Gaussian (Caron and Doucet, 2008)
• normal-Beta-Prime (Bai and Gosh, 2019)

Installation

To install shrinkage from R:

# Install/load R package devtools
install.packages("devtools")
library(devtools)

# Install/load R package beam from github
install_github("gleday/shrinkage")
library(shrinkage)

Vignettes

Vignette describing in details the implemented algorithms:

vignette("Algorithms", package = "shrinkage")

Vignette providing code for fitting models using Stan via the R package rstan:

vignette("Stan", package = "shrinkage")