Dear adelisters,
I'm currently using some kind of Artificial neural network ANN) for
clustering/classification.
Why bother you with such things ?
While developing my program for Bray-Curtis ordination, I built an
artificial data set. It contains 400 sites, disposed on an immaginary +,
200 vertically and 200 horizontaly.
On each 200-group, I tried to simulate two perpendicular gradients, using
gaussian response to gradient for species abundance. The species are 10.
I've tried classical ordination on such a table. The main point I want to
point (!) out is that Self-Organizing Features Maps succed in separating
the 400 sites in two cluster: one for the horizontal axis and the second
for the vertical one.
the code I've used could be ftp'ed (what an awful expression for french
lovers ! I'm sorry.) from:
ftp://cochlea.hut.fi/pub/som_pak
I'll like some feedback from you all.
----------------------------------------
Federico Spinazzi
email: federico@syspr03.disat.unimi.it
----------------------------------------
This archive was generated by hypermail 2b30 : Sat Feb 10 2001 - 10:21:54 MET