horserule: hs – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

hs

Horseshoe regression Gibbs-sampler

Description

Generates posterior samples using the horseshoe prior. The Gibbs sampling method from Makalic E and Schmidt DF (2016). “A Simple Sampler for the Horseshoe Estimator.” IEEE Signal Process. Lett., 23(1), pp. 179–182. is used to generate the posterior samples.

Usage

hs(X, y, niter = 1000, hsplus = F, prior = NULL, thin = 1,
  restricted = 0)

Arguments

`X`	A matrix containing the predictor variables to be used.
`y`	The vector of numeric responses.
`niter`	Number of posterior samples.
`hsplus`	If "hsplus=T" the horseshoe+ extension will be used.
`prior`	Prior for the individual predictors. If all 1 a standard horseshoe model is fit.
`thin`	If > 1 thinning is performed to reduce autocorrelation.
`restricted`	Threshold for restricted Gibbs sampling. In each iteration only coefficients with scale > restricted are updated. Set restricted = 0 for unrestricted Gibbs sampling.

Value

A list containing the posterior samples of the following parameters:

`beta`	Matrix containing the posterior samples for the regression coefficients.
`sigma`	Vector contraining the Posterior samples of the error variance.
`tau`	Vector contraining the Posterior samples of the overall shrinkage.
`lambda`	Matrix containing the posterior samples for the individual shrinkage parameter.

Examples

x = matrix(rnorm(1000), ncol=10)
y = apply(x,1,function(x)sum(x[1:5])+rnorm(1))
hsmod = hs(X=x, y=y, niter=100)

horserule

Flexible Non-Linear Regression with the HorseRule Algorithm

v1.0.0

GPL-3

Authors

Malte Nalenz <malte.nlz@googlemail.com>, Mattias Villani <mattias.villani@liu.se>

Initial release

2018-03-21

hs