clue (version 0.3-2)

ensemble: Cluster Ensembles

Description

Creation and manipulation of cluster ensembles.

Usage

cl_ensemble(..., list = NULL)
as.cl_ensemble(x)
is.cl_ensemble(x)

Arguments

...

R objects representing clusterings or dissimilarities on the same objects.

list

a list of R objects as in ....

x

for as.cl_ensemble, an R object as in ...; for is.cl_ensemble, an arbitrary R object.

Value

• cl_ensemble returns a list of the given clusterings or dissimilarities, with additional class information (see Details).

Details

cl_ensemble creates cluster ensembles, which are realized as lists of clusterings (or dissimilarities) with additional class information, always inheriting from "cl_ensemble". All elements of the ensemble must have the same number of objects.

If all elements are partitions, the ensemble has class "cl_partition_ensemble"; if all elements are dendrograms, it has class "cl_dendrogram_ensemble" and inherits from "cl_hierarchy_ensemble"; if all elements are hierarchies (but not always dendrograms), it has class "cl_hierarchy_ensemble". Note that empty or mixed ensembles cannot be categorized according to the kind of elements they contain, and hence only have class "cl_ensemble".

The list representation makes it possible to use lapply for computations on the individual clusterings in (i.e., the components of) a cluster ensemble.

Available methods for cluster ensembles include those for subscripting, c, rep, and print.

Note that (currently), as.cl_ensemble assumes that unclassed lists represent single clusterings, as this in particular holds true for kmeans in versions of R prior to 2.1.0. This may change eventually.

Examples

d <- dist(USArrests)
hclust_methods <- c("ward", "single", "complete", "average",
                    "mcquitty", "median", "centroid")
hclust_results <- lapply(hclust_methods, function(m) hclust(d, m))
## Now create an ensemble from the results.
hens <- cl_ensemble(list = hclust_results)
names(hens) <- hclust_methods 
hens
## Subscripting.
hens[1 : 3]
## Replication.
rep(hens, 3)
## And continue to analyze the ensemble, e.g.
round(cl_dissimilarity(hens, method = "gamma"), 4)