Simulated 3-dimensional Normal Data Containing 8 Clusters

Description

Cluster means are given by the 8 possible values of (+/-2, +/-2, +/-2) to which standard normal noise was added. True clusters are given by rep(1:8,each =50).

Usage

data(Ysim2)

Format

matrix with 400 rows and 3 columns.

Source

Simulated by
2 * matrix(c(rep(c(1,1,1),50), rep(c(1,1,-1),50), rep(c(1,-1,1),50), rep(c(-1,1,1),50), rep(c(1,-1,-1),50), rep(c(-1,1,-1),50), rep(c(-1,-1,1),50), rep(c(-1,-1,-1),50)), byrow=TRUE, ncol=3) + matrix(rnorm( 400*3),ncol=3)

References

Fritsch, A. and Ickstadt, K. (2009) An improved criterion for clustering based on the posterior similarity matrix, Bayesian Analysis, accepted.

Results