# ex2.draw: MCMC samples of Bayesian cluster model for a simulated... In muschellij2/mcclust.ext: Point estimation and credible balls for Bayesian cluster analysis

## Description

MCMC samples of clusterings from a Dirichlet process scale-location mixture model with normal components fitted to a simulated dataset, see `ex2.data`. True clusters are located at (+/- 2, +/- 2) with a standard deviation of 1, 0.5, 1, and 1.5 in the first, second, third, and fourth quadrant respectively.

## Usage

 `1` ```data(ex2.draw) ```

## Format

The matrix `ex2.draw` has 10,000 rows and 200 columns, with each row representing a MCMC posterior sample of the clustering of the 200 data points contained in `ex2.data`.

## Source

Wade, S. and Ghahramani, Z. (2015) Bayesian cluster analysis: Point estimation and credible balls. Submitted. arXiv:1505.03339.

## Examples

 ``` 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19``` ```data(ex2.data) data(ex2.draw) x=data.frame(ex2.data[,c(1,2)]) cls.true=ex2.data\$cls.true plot(x[,1],x[,2],xlab="x1",ylab="x2") k=max(cls.true) for(l in 2:k){ points(x[cls.true==l,1],x[cls.true==l,2],col=l)} # Find representative partition of posterior psm=comp.psm(ex2.draw) ex2.VI=minVI(psm,ex2.draw,method=("all"),include.greedy=TRUE) summary(ex2.VI) plot(ex2.VI,data=x) # Uncertainty in partition estimate ex2.cb=credibleball(ex2.VI\$cl[1,],ex2.draw) summary(ex2.cb) plot(ex2.cb,data=x) ```

muschellij2/mcclust.ext documentation built on May 26, 2019, 9:36 a.m.