mochoice {mcsm}R Documentation

An MCMC model choice illustration for the linear model

Description

Using a Gibbs sampling strategy of changing one indicator at a time, this function explores the space of models and returns the most likely models among those visited. The data used in this example is swiss, with four explanatory variables.

Usage

mochoice(Niter = 10^4)

Arguments

Niter Number of MCMC iterations

Value

model Sequence of model indicators visited by the MCMC algorithm
top Five most likely models

Note

For more details, see Chapter 3 of Bayesian Core (2007, Springer-Verlag) by J.-M. Marin and C.P. Robert, since the procedure is derived from the developments in this chapter.

Author(s)

Christian P. Robert and George Casella

References

From Chapter 6 of EnteR Monte Carlo Statistical Methods

Examples

mochoice(10^3)

[Package mcsm version 1.0 Index]