Plot Entropy Plots
Plot "entropy plots" to help select the number of classes from a hierarchy of combined clusterings.
entPlot(z, combiM, abc = c("standard", "normalized"), reg = 2, ...)z |
A matrix whose |
combiM |
A list of "combining matrices" (as provided by |
abc |
Choose one or more of: "standard", "normalized", to specify whether the number of observations involved in each combining step should be taken into account to scale the plots or not. |
reg |
The number of parts of the piecewise linear regression for the entropy plots. Choose one or more of: 2 (for 1 change-point), 3 (for 2 change-points). |
... |
Other graphical arguments to be passed to the plot functions. |
Please see the article cited in the references for more details. A clear elbow in the "entropy plot" should suggest the user to consider the corresponding number(s) of class(es).
if abc = "standard", plots the entropy against the number of clusters and the difference between the entropy of successive combined solutions against the number of clusters.
if abc = "normalized", plots the entropy against the cumulated number of observations involved in the successive combining steps and the difference between the entropy of successive combined solutions divided by the number of observations involved in the corresponding combining step against the number of clusters.
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):332-353.
data(Baudry_etal_2010_JCGS_examples) # run Mclust to get the MclustOutput output <- clustCombi(data = ex4.2, modelNames = "VII") entPlot(output$MclustOutput$z, output$combiM, reg = c(2,3)) # legend: in red, the single-change-point piecewise linear regression; # in blue, the two-change-point piecewise linear regression.
Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.