[ncl-talk] k-means clustering

Dennis Shea shea at ucar.edu
Wed Aug 17 10:30:42 MDT 2016


There are multiple methods for determining the 'optimal' number of clusters.
The issue is ... which one?

A statistician was asked for suggestions. In classic 'statistician-speak',
he rambled on about assorted methods to estimate the optimal number of
clusters. It was a bit confusing. Esentially, his answer was 'try a bunch
and see what you like.'

NCL is not 'R' or Matlab or .... NCL can not provide and support multiple
methods. NCL would like to include one optimal cluster function. BUT don't
have the time to really look at this issue. However, if you find something
you like, we will see what we can do.uses clustering ("classification")
methods.

===

Several papers, which uses classification/clustering are:

http://www.cgd.ucar.edu/staff/cdeser/docs/cassou.natlregimes.jclim04.pdf
**See Part 2 (Classification)**

http://www.nature.com/nature/journal/v455/n7212/full/nature07286.html

ftp://ftp.cerfacs.fr/pub/globc/exchanges/cassou/DCLIM/
cassou_et_al_climdyn_regime_ocean_part1.pdf


HTH
D



On Wed, Aug 17, 2016 at 7:02 AM, Carl Schreck <cjschrec at ncsu.edu> wrote:

> I'm just starting to play around with the new kmeans_as136
> <http://www.ncl.ucar.edu/Document/Functions/Built-in/kmeans_as136.shtml> function.
> I'm very much a novice at clustering and am wondering if anyone has used
> this function and has guidance on how best to choose the number of
> clusters.
>
> Most of what I'm finding is similar to this wikipedia article
> <https://en.wikipedia.org/wiki/Determining_the_number_of_clusters_in_a_data_set>.
> I think the elbow method is the only one that we can do with the output
> from kmeans_as136, but I'm not exactly sure how to relate ss2 to %variance.
>
> I'm sure this is simple and I've just not read up enough on it, but I
> figured I'd ask you guys. Thanks!
>
> Carl
>
> --
>
> <http://www.cicsnc.org/>
> <http://www.researcherid.com/rid/B-8711-2011>
> * Cyclone <http://www.cyclonecenter.org/>*Center.org
> <http://www.cyclonecenter.org/>   *Carl J. Schreck III, PhD*
> *Research Scholar / Research **Assistant Professor*
> Cooperative Institute for Climate and Satellites NC (CICS
> <http://www.cicsnc.org/>*-NC) <http://www.cicsnc.org/> /*
> Department of Marine, Earth, and Atmospheric Sciences (MEAS)
> <https://meas.sciences.ncsu.edu/>
> North Carolina State University <http://ncsu.edu/>
> NOAA National Centers for Environmental Information (NCEI)
> <http://ncei.noaa.gov/>
> 151 Patton Ave, Asheville, NC 28801
> e: carl_schreck at ncsu.edu
> o: +1 828 257 3140
> c: +1 828 484 1702
> Publications
> <http://scholar.google.com/citations?hl=en&user=th8ONEcAAAAJ&view_op=list_works&sortby=pubdate>
> monitor.cicsnc.org/mjo
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20160817/d9767e5a/attachment.html 


More information about the ncl-talk mailing list