cutree
Create Groups from Hierarchical Clustering
Description
Returns a vector of group numbers for the observations that were clustered.
Specify either the number of groups desired or a clustering height.
Usage
cutree(tree, k = NULL, h = NULL)
Arguments
| tree |
a hierarchical clustering tree structure of class "hclust".
It is commonly the output of the hclust function.
|
| k |
the desired number of groups. Must be between 1 and the row number
of the "merge" component of tree.
If k is a vector of integers, the output will be a matrix
with a column for each value in k.
The default is NULL.
Either k or h must be non-NULL, if both are
non-NULL, then k is used and h is ignored.
|
| h |
the height at which to cut tree to produce the groups.
Groups are defined by the structure of the tree above the cut.
If h is a vector of numbers, the output will be a matrix
with a column for each value in h.
The default is NULL.
Either k or h must be non-NULL, if both are
non-NULL then k is used and h is ignored.
|
Value
For scalar k or h, cutree returns a vector
with as many elements as there are leaves in the tree.
The ith element of the vector gives the
group number to which individual i is assigned.
For vector k or h (of length more than 1),
cutree returns a matrix with a column of group numbers
for each value in k or h.
Differences between Spotfire Enterprise Runtime for R and Open-source R
The group numbering might be different between Spotfire Enterprise Runtime for R and open-source R, but the
group memberships are the same.
In open-source R, when k is NULL, and if
the "height" component of tree is not sorted,
cutree is stopped. Spotfire Enterprise Runtime for R sorts the "height" component to
ensure "height" is ordered as increasing.
References
Becker, R. A., Chambers, J. M., and Wilks, A. R. 1988. The New S Language: A Programming Environment for Data Analysis and Graphics. Pacific Grove, CA: Wadsworth & Brooks/Cole Advanced Books and Software.
See Also
Examples
hc <- hclust(dist(c(12, 2, 5, 4, 3)))
cutree(hc, k=5)
vote.cl <- hclust(dist(Sdatasets::votes.repub))
group <- cutree(vote.cl, k=5) # produce 5 groups
Sdatasets::state.name[group==1]