Just a side remark : "any" clustering method in R is pretty broad. Actually, k-means clustering is a completely different animal than distance-based neighbour-joining methods. You should be a bit more specific about your research
I wouldn't worry too much about the scientific practice, as cluster analysis is inherently a heuristic method, and thus not really well defined in an inference framework. One way you could proceed, is work with the dissimilarity matrix. If the groups you want to study are well defined (e.g. young and old patients, treatment and no treatment), you can test whether the distances within each group are smaller than outside the group. A glm procedure should be able to give you some directions at least. The only thing I don't know, is how these distances are distributed, and hence which link function and/or base functions are appropriate. Kind regards Joris On Wed, Mar 18, 2009 at 2:43 PM, Carlos J. Gil Bellosta < c...@datanalytics.com> wrote: > Dear R-helpers, > > I am writing to the list in order to inquire whether there exists any > R package or program that will help me "describe" clusters. > > The situation is as follows: > > 1) I create some clusters (say, with any clustering method in R). > > 2) I want to "describe" and assign some kind of "label" to each of them. > > In order to "label" each cluster, I want to compare the distribution > of the variables in each cluster with respect to the distribution of > the variables in the original dataset. I would like to do it > graphically, if possible. In this way I could review this output and > say: this cluster corresponds to, say, "older patients who were not > treated before", etc. > > I am aware this is not sound scientific practice, but I am asked to do > something like that. I have some ideas about how to do it, but I would > like to know if I am walking on a well trodden path. > > Best regards, > > Carlos J. Gil Bellosta > http://www.datanalytics.com > > ______________________________________________ > R-help@r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide > http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. > [[alternative HTML version deleted]] ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
