[R] "FANNY" function in R package "cluster"

From: Aamir M <intuitionist_at_gmail.com>
Date: Tue 31 May 2005 - 05:57:29 EST

Dear All,

I am attempting to use the FANNY fuzzy clustering function in R
(Kaufman & Rousseeuw, 1990), found in the "cluster" package. I have
run into a variety of difficulties; the two most crucial difficulties are enumerated below.

1. Where is the 'm' parameter in FANNY? In _Finding Groups in Data: An Introduction to Cluster Analysis_
(1990) by Kaufman & Rousseeuw, the authors discuss the FANNY
algorithm. On pages 189-190, they attempt to demonstrate an equivalence between Fuzzy c-Means (FCM) and FANNY. In doing so they, appear to be assuming that the value of the 'm' parameter in FCM (a measure of the fuzziness) is fixed at m=2. Although this is how FCM was originally formulated, it eventually became apparent that m should be a user-specified parameter and not a fixed value of m=2. My question, then, is twofold. Firstly, am I correct in saying that Kaufman & Rousseeuw have assumed m=2 in their formulation of Fuzzy c-Means and FANNY? Secondly, is it possible to modify the FANNY algorithm to allow user-specification of the m (fuzziness) parameter?
2. What do I do with training data? Is there any way to use FANNY for assigning clustering membership values to new, test data? In Fuzzy c-Means, new data is compared to the cluster centers in order to assign clustering membership values to the test data. However, in FANNY these centers do not exist. Is there, then, any way to compute the FANNY clustering membership values of a test data point without affecting the clustering membership values of the training data? Perhaps there are enough conditions to use the objective function as a way of computing the membership values of the test data?

Aamir M
University of Toronto

R-help@stat.math.ethz.ch mailing list