It's in PDL-Stats http://pdl-stats.sourceforge.net/

Matlab statistics toolbox has a version of kmeans
http://www.mathworks.com/access/helpdesk/help/toolbox/stats/kmeans.html

Matlab kmeans has an optional "online update" phase which is not
implemented in PDL::Stats::Kmeans. Easy to implement but it's way too
time-consuming / impractical for processing large data sets. A better
way is to use PCA to seed Kmeans analysis, imho, which is pretty easy
to do with PDL::Stats. I don't think Matlab kmeans threads.


Maggie


On Sun, Nov 8, 2009 at 7:30 AM, Gabor Szabo <[email protected]> wrote:
> Maggie,
>
> On Wed, Nov 4, 2009 at 4:42 PM, Maggie X <[email protected]> wrote:
>> I was absolutely awed by PDL when I was able to write a threaded version of
>> K-means cluster analysis http://en.wikipedia.org/wiki/K-means_clustering
>> This is not your a + b vector operation, but a whole analysis, threaded if
>> you feed it extra data dimensions. I tested it with up to 4D data and
>> various scenarios. It's so overly powerful that I'm not sure what kind of
>> scenario needs it :P Maybe gene analysis, but that's not my kind of data.
>>
>> And all that from a person who barely knows C.
>>
>>
>> Maggie
>>
>
> Is that code available under and open source license?
> If not yet, would it be possible to make it available?
>
> Could you give a bit more detail about where the K means clustering fits in?
> Eg. is it part of one of the Matlab toolboxes?
>
> Gabor
>

_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl

Reply via email to