posteriorMCMC.nl {BMAmevt}

MCMC posterior samplers for the pairwise beta and the negative logistic models.

Description

The functions generate parameters samples approximating the posterior distribution in the PB model or the NL model.

Usage

posteriorMCMC.nl(Nsim, dat, Hpar, MCpar, ...)

posteriorMCMC.pb(Nsim, dat, Hpar, MCpar, ...)

Arguments

Nsim

Total number of iterations to perform.

dat

An angular data set, e.g., constructed by cons.angular.dat: A matrix which rows are the Cartesian coordinates of points on the unit simplex (summing to one).

Hpar

A list containing Hyper-parameters to be passed to prior.

MCpar

A list containing MCMC tuning parameters to be passed to proposal.

...

Additional arguments to be passed to posteriorMCMC instead of their default values (must not contain any of "prior", "likelihood", "proposal", "name.model" or "class").

Details

The two functions are wrappers simplifying the use of posteriorMCMC for the two models implemented in this package.

Value

an object with class attributes "postsample" and "PBNLpostsample": The posterior sample and some statistics as returned by function posteriorMCMC

Note

For the Leeds data set, and for simulated data sets with similar features, setting Nsim=50e+3 and Nbin=15e+3 is enough (possibly too much), with respect to the Heidelberger and Welch tests implemented in heidel.diag.

See Also

posteriorMCMC

Examples

## Not run: 
data(Leeds)
data(pb.Hpar)
data(pb.MCpar)
data(nl.Hpar)
data(nl.MCpar)
pPB <- posteriorMCMC.pb(Nsim=5e+3, dat=Leeds, Hpar=pb.Hpar,
MCpar=pb.MCpar)

dim(pPB[1])
pPB[-(1:3)]

pNL <- posteriorMCMC.nl(Nsim=5e+3, dat=Leeds, Hpar=nl.Hpar,
MCpar=nl.MCpar)

dim(pNL[1])
pNL[-(1:3)]

## End(Not run)