Draw an ICLUST hierarchical cluster structure diagram
Given a cluster structure determined by ICLUST, create a graphic structural diagram using graphic functions in the psych package To create dot code to describe the ICLUST output with more precision, use ICLUST.graph. If Rgraphviz has been successfully installed, the alternative is to use ICLUST.rgraph.
iclust.diagram(ic, labels = NULL, short = FALSE, digits = 2, cex = NULL, min.size = NULL,
e.size =1,colors=c("black","blue"),
main = "ICLUST diagram",cluster.names=NULL,marg=c(.5,.5,1.5,.5))ic |
Output from ICLUST |
labels |
labels for variables (if not specified as rownames in the ICLUST output |
short |
if short=TRUE, variable names are replaced with Vn |
digits |
Round the path coefficients to digits accuracy |
cex |
The standard graphic control parameter for font size modifications. This can be used to make the labels bigger or smaller than the default values. |
min.size |
Don't provide statistics for clusters less than min.size |
e.size |
size of the ellipses with the cluster statistics. |
colors |
postive and negative |
main |
The main graphic title |
cluster.names |
Normally, clusters are named sequentially C1 ... Cn. If cluster.names are specified, then these values will be used instead. |
marg |
Sets the margins to be narrower than the default values. Resets them upon return |
iclust.diagram provides most of the power of ICLUST.rgraph without the difficulties involved in installing Rgraphviz. It is called automatically from ICLUST.
Following a request by Michael Kubovy, cluster.names may be specified to replace the normal C1 ... Cn names.
If access to a dot language graphics program is available, it is probably better to use the iclust.graph function to get dot output for offline editing.
Graphical output summarizing the hierarchical cluster structure. The graph is drawn using the diagram functions (e.g., dia.curve, dia.arrow, dia.rect, dia.ellipse ) created as a work around to Rgraphviz.
Suggestions for improving the graphic output are welcome.
William Revelle
Revelle, W. Hierarchical Cluster Analysis and the Internal Structure of Tests. Multivariate Behavioral Research, 1979, 14, 57-74.
v9 <- sim.hierarchical()
v9c <- ICLUST(v9)
test.data <- Harman74.cor$cov
ic.out <- ICLUST(test.data)
#now show how to relabel clusters
ic.bfi <- iclust(psychTools::bfi[1:25],beta=3) #find the clusters
cluster.names <- rownames(ic.bfi$results) #get the old names
#change the names to the desired ones
cluster.names[c(16,19,18,15,20)] <- c("Neuroticism","Extra-Open","Agreeableness",
"Conscientiousness","Open")
#now show the new names
iclust.diagram(ic.bfi,cluster.names=cluster.names,min.size=4,e.size=1.75)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.