cutree-methods: Cut Out Clustering Level from Cluster Hierarchy

In apcluster: Affinity Propagation Clustering

Cut Out Clustering Level from Cluster Hierarchy

Description

Cut out a clustering level from a cluster hierarchy

Usage

## S4 method for signature 'AggExResult'
cutree(tree, k, h)
## S4 method for signature 'APResult'
cutree(tree, k, h)

Arguments

tree	an object of class AggExResult containing a cluster hierarchy; may also be an object of class APResult
k	the level (i.e. the number of clusters) to be selected
h	alternatively, the level can be selected by specifying a cut-off for the merging objective

Details

The function cutree extracts a clustering level from a cluster hierarchy stored in an AggExResult object. Which level is selected can be determined by one of the two arguments k and h (see above). If both k and h are specified, k overrides h. This is done largely analogous to the standard function cutree. The differences are (1) that only one level can be extracted at a time and (2) that an ExClust is returned instead of an index list.

The function cutree may further be used to convert an APResult object into an ExClust object. In this case, the arguments k and h are ignored.

Value

returns an object of class ExClust

Author(s)

Ulrich Bodenhofer and Andreas Kothmeier

References

https://github.com/UBod/apcluster

Bodenhofer, U., Kothmeier, A., and Hochreiter, S. (2011) APCluster: an R package for affinity propagation clustering. Bioinformatics 27, 2463-2464. DOI: \Sexpr[results=rd]{tools:::Rd_expr_doi("10.1093/bioinformatics/btr406")}.

See Also

AggExResult, ExClust

Examples

## create two simple clusters
x <- c(1, 2, 3, 7, 8, 9)
names(x) <- c("a", "b", "c", "d", "e", "f")

## compute similarity matrix (negative squared distance)
sim <- negDistMat(x, r=2)

## run affinity propagation
aggres <- aggExCluster(sim)

## show details of clustering results
show(aggres)

## retrieve clustering with 2 clusters
cutree(aggres, 2)

## retrieve clustering with cut-off h=-1
cutree(aggres, h=-1)

apcluster documentation built on May 29, 2024, 2:25 a.m.