Compute a kmedoids partition of a dissimilarity object.
1  kmedoids(x, k)

x 
a dissimilarity object inheriting from class

k 
an integer giving the number of classes to be used in the partition. 
Let d denote the pairwise objecttoobject dissimilarity matrix
corresponding to x
. A kmedoids partition of x
is
defined as a partition of the numbers from 1 to n, the number of
objects in x
, into k classes C_1, …, C_k such
that the criterion function
L = ∑_l \min_{j \in C_l} ∑_{i \in C_l} d_{ij}
is minimized.
This is an NPhard optimization problem. PAM (Partitioning Around
Medoids, see Kaufman & Rousseeuw (1990), Chapter 2) is a very popular
heuristic for obtaining optimal kmedoids partitions, and
provided by pam
in package cluster.
kmedoids
is an exact algorithm based on a binary linear
programming formulation of the optimization problem (e.g., Gordon &
Vichi (1998), [P4']), using lp
from package
lpSolve as solver. Depending on available hardware resources
(the number of constraints of the program is of the order n^2),
it may not be possible to obtain a solution.
An object of class "kmedoids"
representing the obtained
partition, which is a list with the following components.
cluster 
the class ids of the partition. 
medoid_ids 
the indices of the medoids. 
criterion 
the value of the criterion function of the partition. 
L. Kaufman and P. J. Rousseeuw (1990). Finding Groups in Data: An Introduction to Cluster Analysis. Wiley, New York.
A. D. Gordon and M. Vichi (1998). Partitions of partitions. Journal of Classification, 15, 265–285. doi: 10.1007/s003579900034.
Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.
Please suggest features or report bugs with the GitHub issue tracker.
All documentation is copyright its authors; we didn't write any of that.