Make all estimated clusters agree with a pivot allocation

Description

Given a pivot allocation vector, a set of simulated allocations and a set of permutations from different relabelling algorithms, this function relabels the permutations so that all methods maximize their similarity with the pivot. This is helpful when comparing different different label switching algorithms.

Usage

1
compare.clust(pivot.clust,perms,z,K)

Arguments

pivot.clust

a pivot allocation vector of the n observations among the K clusters.

perms

a list containing f permutation arrays, as returned by label.switching function.

z

a set of simulated allocation arrays.

K

number of mixture components

Value

similarity

(f+1) K\times (f+1) matrix containing the similarity coefficient of the resulting clusters.

clusters

f\times n array of single best clusterings, relabelled in order to maximize their similarity with pivot.clust.

permutations

releaballed permutations.

Author(s)

Panagiotis Papastamoulis

See Also

label.switching