Sorry for the spam.
I'm trying to do k-means or k-modes clustering in SAS with given type of dissimilarity. As I have noticed that proc fastclus only accepts raw data or maybe coordinates, but the default distance is given by Euclidean method. For proc cluster, although it can deal with dissimilarity matrix, it only gives us hierarchically clustering. So my question is if there is anything that lies between them. I mean which can take both dissimilarity matrix as input and do k-modes / k-means clustering.
Btw. functions in R are also acceptable. Thank you very much.
Sincerely,
Danhua
_________________________________________________________________
Search on the go: Try Windows Live(tm) Search for Mobile beta
http://www1.imagine-msn.com/minisites/mobile/Default.aspx?locale=en-us
|