R Graphical Manual

Browse All

Last data update: 2014.03.03

R: Stochastic matching pursuit for variable selection.

CompWiseGibbsSMP

R Documentation

Stochastic matching pursuit for variable selection.

Description

Perform MCMC procedure to generate the posterior samples to estimate posterior quantities of interest in Bayesian variable selection using stochastic matching pursuit approach (SMP).

Usage

CompWiseGibbsSMP(Y, X, beta.value, r, tau2, rho, sigma2, nu, lambda, 
num.of.inner.iter, num.of.iteration, MCSE.Sigma2.Given)

Arguments

`Y`	vector of observations of length n.
`X`	design matrix of dimension n \times p.
`beta.value`	Initial values of regression coefficients, β.
`r`	Initial values of indicator variables for individual regressors.
`tau2`	Variance in the prior distribution for regression coefficients.
`rho`	Prior probability including a variable.
`sigma2`	Initial value of σ^2.
`nu`	Given value in the prior distribution of σ^2.
`lambda`	Given value in the prior distribution of σ^2.
`num.of.inner.iter`	The number of iterations before sampling σ^2.
`num.of.iteration`	The number of iterations to be runned for sparse group variable selection.
`MCSE.Sigma2.Given`	Prespecified value which is used to stop simulating samples when the MCSE of estimate of σ^2 less then given values.

Value

A list is returned with posterior samples of regression coefficients, β, variance σ^2, binary variables, γ, the number of iterations performed, and the time in second required for the run.

Examples