kml3d (version 0.6)

plot(ListCLustering): ~ Function: plot for ListClustering ~

Description

This function graphically displays the quality criterion of all the Clustering of a ListClustering object.

Usage

## S3 method for class 'ListClustering,missing':
plot(x, criterion=x['criterionActif'],nbCriterion = 100, standardized = TRUE)

Arguments

x

[ListClustering]: object whose quality criterion should be displayed.

criterion

[character] or [vector(character)]: name of the criterion(s) to plot (default= "calinksi").

nbCriterion

[numeric]: set the maximum number of criterion to display for a specific clusters number. When several criterions are plot, nbCriterion is set to 1 (default=100).

standardized

[logical]: some criterion sould be maximized, some other should be minimized ; some criterion take big values, some take small values. If standardized is TRUE, all the criterion are scale to be in [0,1] and / or ar

Value

• No value are return. A graph is printed.

Author(s)

Christophe Genolini INSERM U669 / PSIGIAM: Paris Sud Innovation Group in Adolescent Mental Health Modal'X / Universite Paris Ouest-Nanterre- La Defense Contact author : [e-mail]

Details

This function display graphically the quality criterion (probably to decide the best clusters' number). It can either display all the value find for a specific criterion ; this is useful to see the consistency of the result : is the best clusterization obtain several time or only one ? It can also display the values find by several criterions : this helps to compare the best cluster's number find by various ways.

References

Article "KmL: K-means for Longitudinal Data", in Computational Statistics, Volume 25, Issue 2 (2010), Page 317. Web site: http://christophe.genolini.free.fr/kml

Examples

##############
### Preparing data

### Longdata
traj <- as.longData(array(rnorm(30*5*2),dim=c(30,5,2)))

### Some clustering
part2 <- partition(rep(c("A","B","A"),time=10),nbClusters=2)
clust2 <- clustering(traj,part2)
part2b <- partition(rep(c("A","B","A"),time=10),nbClusters=2)
clust2b <- clustering(traj,part2b)

part3 <- partition(rep(c("A","B","C"),time=10),nbClusters=3)
clust3 <- clustering(traj,part3)
part3b <- partition(rep(c("A","B","C","B","C"),time=6),nbClusters=3)
clust3b <- clustering(traj,part3b)

part4 <- partition(rep(c("A","B","A","C","D"),time=6),nbClusters=4)
clust4 <- clustering(traj,part4)
part4b <- partition(rep(c("A","B","A","C","D"),time=6),nbClusters=4)
clust4b <- clustering(traj,part4b)

################
### ListClustering
listClus <- listClustering()
listClus['criterionActif'] <-"calinski"
plot(listClus)

listClus["add"] <- clust2
listClus["add"] <- clust2b
listClus["add"] <- clust3
listClus["add"] <- clust3b
listClus["add"] <- clust4
listClus["add"] <- clust4b
plot(listClus)

### Plot a single criterion
ordered(listClus)
plot(listClus)

### Comparing several criterion
plot(listClus,nbCriterion=1,criterion=c("calinski","test"),standardized=FALSE)
plot(listClus,nbCriterion=1,criterion=c("calinski","test"))