gclust.boxstats {sna}

Plot Statistics Associated with Graph Clusters

Description

gclust.boxstats creates side-by-side boxplots of graph statistics based on a hierarchical clustering of networks (cut into k sets).

Usage

gclust.boxstats(h, k, meas, ...)

Arguments

h

an hclust object, presumably formed by clustering a set of structural distances.

k

the number of groups to evaluate.

meas

a vector of length equal to the number of graphs in h, containing a GLI to be evaluated.

...

additional parameters to boxplot.

Details

gclust.boxstats simply takes the hclust object in h, applies cutree to form k groups, and then uses boxplot on the distribution of meas by group. This can be quite handy for assessing graph clusters.

Value

None

Note

Actually, this function will work with any hclust object and measure matrix; the data need not originate with social networks. For this reason, the clever may also employ this function in conjunction with sedist or equiv.clust to plot NLIs against clusters of positions within a graph.

Author(s)

Carter T. Butts [e-mail]

References

Butts, C.T., and Carley, K.M. (2001). “Multivariate Methods for Interstructural Analysis.” CASOS working paper, Carnegie Mellon University.

See Also

gclust.centralgraph, gdist.plotdiff, gdist.plotstats

Examples

#Create some random graphs
g<-rgraph(10,20,tprob=c(rbeta(10,15,2),rbeta(10,2,15)))

#Find the Hamming distances between them
g.h<-hdist(g)

#Cluster the graphs via their Hamming distances
g.c<-hclust(as.dist(g.h))

#Now display boxplots of density by cluster for a two cluster solution
gclust.boxstats(g.c,2,gden(g))