Fog Creek Software
Discussion Board

grouping data

I have n nodes.  Each node has a 'score' against every other node (score for node 20 vs 5 is not necessarily the score for 5 vs 20 though).

How would I detect groups of nodes with similiar scores relative to everything else?

I've started investigating self-organising-maps (SOMs), which would give me a 2d grid which I can then work out some kind of clustering from somehow.

Are there any other techniques you can introduce me to?

i like i
Wednesday, August 27, 2003

*  Recent Topics

*  Fog Creek Home