comp.lang.idl-pvwave archive
Messages from Usenet group comp.lang.idl-pvwave, compiled by Paulo Penteado

Home » Public Forums » archive » clustering
Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend 
Return to the default flat view Create a new topic Submit Reply
Re: clustering [message #54963 is a reply to message #54850] Mon, 23 July 2007 09:08 Go to previous messageGo to previous message
little davey is currently offline  little davey
Messages: 5
Registered: June 2007
Junior Member
On Jul 17, 10:36 am, nivedita.raghun...@gmail.com wrote:
> Hi all,
>
> I am trying to cluster a (7 x n) array with n_clusters=5. Visually I
> can see 5 distinct clusters, but when I do clust_wts the cluster
> centroids don't end up right. No matter what options I give, clust_wts
> refuses to find the clusters.
>
> Any idea on whats going on ?
>
> Thanks,
> Nivedita

I seem to recall that "clust_wts.pro" just doesn't work. Because of
"work" I don't have time to dig into it, but I have glanced at a file
I have called "myclust_wts.pro" that I copied from the IDL directory.
I recall having stepped through the program, and I found that, oh, I
think it didn't "converge" or something like that. I hate saying this
without having the decency to do my homework, but my memory is that
CLUST_WTS just does not work. The algorithm I thought was not
adequate. I'm looking at where it says:

;Normalized uniformly random cluster weights.
;;Weights = RANDOMU(SEED, N_Variables, N_Clusters) + Zero
av1 = average(array[0,*])
av2 = average(array[1,*])
Weights = RANDOMU(SEED, N_Variables, N_Clusters) + Zero
for k = 0L, N_Clusters-1 do Weights[*,k] = $
(Weights[*,k] / TOTAL(Weights[*,k])) * Variable_Wts

I recall something like the weights and the data simply didn't have
the same data range, so the algorithm failed. Sorry if I'm totally
wrong.

-- Dave K --
[Message index]
 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: Re: Move data to structure fields without copying
Next Topic: Annotate Uncertainty [Upper & Lower] on a plot?

-=] Back to Top [=-
[ Syndicate this forum (XML) ] [ RSS ] [ PDF ]

Current Time: Sat Nov 29 13:54:14 PST 2025

Total time taken to generate the page: 0.48347 seconds