mclust (version 1.1-7)

awe: Approximate weight of evidence for model-based hierarchical clustering.

Description

Computes a Bayesian criterion for assessing the number of clusters present in the data.

Usage

awe(tree, data)

Arguments

tree

an mhtree object.

data

the data used to produce the mhtree object.

Value

• the approximate weight of evidence for each possible stage of merging.

NOTES

Since mhtree allows stopping and starting at any stage, the result will contain NAs for those stages that have been eliminated. If you scaled your data before using mhtree, be sure to use the same scaling when supplying the data to awe.

References

J. D. Banfield and A. E. Raftery, Model-based Gaussian and non-Gaussian Clustering, Biometrics,49:803-821 (1993).

See Also

mhtree, loglik

Examples

data(iris)
iris.m _ iris[,1:4]
awe.val <- awe(mhtree(iris.m), iris.m)
plot(awe.val)