clus2norMix {nor1mix}

Transform Clustering / Grouping to Normal Mixture

Description

Simple transformation of a clustering or grouping to a normal mixture object (class "norMix", see, norMix.

Usage

clus2norMix(gr, x, name = deparse(sys.call()))

Arguments

gr

a grouping/clustering vector with values in \{1,..,K\}; possibly a factor.

x

numeric vector of (original) data (of the same length as gr).

name

name for norMix() object; constructed from the call by default.

Value

A call to norMix() with (mu, sig2, w) set to the empirical values of the groups (as defined by split(x,gr).

Note

Via this function, any simple clustering algorithm (such pam) can be used as simple mixture model fitting procedure.

Author(s)

Martin Maechler, Dec. 2007

See Also

norMix; further pam() (or clara()) from package cluster for sensible clusterings.

Examples

x9 <- rnorMix(500, MW.nm9)
require("cluster")
pxc <- pam(x9, k=3)
plot(pxc, which = 2)# silhouette

(nm.p9 <- clus2norMix(pxc$clustering, x9))
plot(nm.p9, p.norm=FALSE)
lines(MW.nm9, col="thistle")