Plot classifications corresponding to successive combined solutions.
1 
data 
The data. 
z 
A matrix whose [i,k]th entry is the probability that observation i in the data belongs to the kth class, for the initial solution (ie before any combining). Typically, the one returned by 
combiM 
A "combining matrix" (as provided by 
... 
Other arguments to be passed to the 
Plot the classifications obtained by MAP from the matrix t(combiM %*% t(z))
, which is the matrix whose [i,k]th entry is the probability that observation i in the data belongs to the kth class, according to the combined solution obtained by merging (according to combiM
) the initial solution described by z
.
J.P. Baudry, A. E. Raftery, L. Scrucca
J.P. Baudry, A. E. Raftery, G. Celeux, K. Lo and R. Gottardo (2010). Combining mixture components for clustering. Journal of Computational and Graphical Statistics, 19(2):332353.
clustCombi
, combMat
, clustCombi
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29  data(Baudry_etal_2010_JCGS_examples)
MclustOutput < Mclust(ex4.1)
MclustOutput$G # Mclust/BIC selected 6 classes
par(mfrow=c(2,2))
combiM0 = diag(6) # is the identity matrix
# no merging: plot the initial solution, given by z
combiPlot(ex4.1, MclustOutput$z, combiM0, cex = 3)
title("No combining")
combiM1 = combMat(6, 1, 2) # let's merge classes labeled 1 and 2
combiM1
combiPlot(ex4.1, MclustOutput$z, combiM1)
title("Combine 1 and 2")
# let's merge classes labeled 1 and 2, and then components labeled (in this
# new 5classes combined solution...) 1 and 2
combiM2 = combMat(5, 1, 2) %*% combMat(6, 1, 2)
combiM2
combiPlot(ex4.1, MclustOutput$z, combiM2)
title("Combine 1, 2 and then 1 and 2 again")
plot(0,0,type="n", xlab = "", ylab = "", axes = FALSE)
legend("center", legend = 1:6,
col = mclust.options("classPlotColors"),
pch = mclust.options("classPlotSymbols"),
title = "Class labels:")

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.
All documentation is copyright its authors; we didn't write any of that.