mixer (version 1.8)

plot.mixer: Plot of mixer object

Description

plot.mixer can display five kinds of figure: model selection criterion curve, the adjacency matrix map, the degree distribution histogram, the connectivity matrix graph and the adjacency matrix graph. By default the four first plots are displayed.

Usage

"plot"(x, q=NULL, frame=1:4, classes=NULL,
classes.col=NULL, quantile.val=0.1, ...)

Arguments

x

a mixer object (output of the mixer function).

q

the q-class model to display. By default, the q is set to the value which maximizes the criterion (see frame 1).

frame

a vector of frame numbers to display (5 kinds of plots, see details section for more).

classes

an external classification used for frame 4 (pie chart): vector as factor of node elements (the number of external class levels corresponds to the number of levels).

classes.col

a vector of user colors used to identify the classes.

quantile.val

filters the connectivity matrix values (Pis) in frame 4. Display the upper part (specified by quantile.val) of the distribution.

...

further graphical arguments.

Details

Frame values:

1

criterion (ICL or ILvb) versus the number of classes (see mixer).

2

adjacency matrix reorganized according to the estimated partition for a given number of classes q.

3

degree distribution (histogram) and theoretical degree distribution (blue curve) computed from the q-class model parameters (alphas, Pis).

4

matrix connectivity between classes (Pis) given a number of classes q. The thickest edges identify the highest values of the connectivity probabilities and the largest nodes point out the most populated classes.

Providing external classes (see classes argument) each node displays a pie chart pointing out the classification relevance.

5

graph display of the adjacency matrix.

See Also

mixer, getModel

Examples

#
#  Simple example : display the 4 frames for the best class number estimation
#
g <- graph.affiliation(n=100,c(1/3,1/3,1/3),0.8,0.2)
xout <- mixer(g$x,qmin=2,qmax=6)
## Not run: plot(xout)

#
#  Display the same for 4 classes with no filtering
#
## Not run:  plot(xout, q=4, quantile.val=0) 

#
#  Display a pie chart for 4 classes
#
data(blog)
xout <- mixer(x=blog$links,qmin=2,qmax=12)
#  Unconnected nodes have been removed by mixer.
#  xout$map contains the mapping from connected nodes to the whole set 
ext.classes <-  blog$politicalParty
## Not run:  plot( xout, frame=4, classes=ext.classes )