mixture.gibbs: Runs the Gibbs sampler and returns samples from the posterior...

In drvalle1/EcoCluster: Bayesian Clustering using Truncated Stick-Breaking priors

Description

Runs the Gibbs sampler and returns samples from the posterior distribution

Usage

mixture.gibbs(dat, ngroup, nl, ngibbs, burnin, a.prior, b.prior)

Arguments

dat	this matrix has L rows (locations) and S columns (species) and contains the presence-absence data (i.e., number of times a given species was observed at a given location)
ngroup	maximum number of location groups (K)
nl	this vector has L elements (locations) and contains the number of observation opportunities at each location
ngibbs	number of Gibbs sampler iterations
burnin	number of iterations to discard as burn-in
a.prior	'a' parameter for prior beta distribution
b.prior	'b' parameter for prior beta distribution

Value

this function returns a list containing several matrices. These matrices have ngibbs-burnin rows and contain samples from the posterior distribution for:

• phi: probability of observing each species in each group

• theta: probability of each location group

• logl: log-likelihood

• z: cluster assignment of each location

• gamma: TSB prior parameter

drvalle1/EcoCluster documentation built on Jan. 25, 2022, 12:46 a.m.