In addition to Art's good suggestions, you could obtain coordinate
data by applying multidimensional scaling (MDS) to your 1000 x 1000
matrix of pairwise similarities; then you could apply, for example, k-
means clustering to the 'recovered' coordinate data.
This is a fairly common strategy, I believe.
You could use metric MDS to improve computational efficiency, if
necessary. Then it's not much more computation-intensive than PCA of
a 1000 x 1000 matrix.
HTH
John Uebersax PhD
http://www.satyagraha.com
On Jun 12, 9:42=A0am, Sengly <Sengly.H...@[EMAIL PROTECTED]
> wrote:
> I have browse through various methods such as hierarchy, k-means,
> scaling dimension, etc. I really like k-means method but the problem
> is that I don't have points (and their coordinates) in space but
> rather their similarity.
|
