[STATSGRASS] Similar Areas Cluster analysis

[Roger Bivand]

On Thu, 1 Mar 2007, Martin Wegmann wrote:

> Hello Andi, 
> 
> have a look at http://casoilresource.lawr.ucdavis.edu/drupal/node/221
> that might be what you are looking for. 

Yes, the clara() function in the cluster standard package is good for
large data sets, for continuous variables. You will need to watch the fact
that many of your variables are categorical, though - look at the
supporting documentation for the package. There is also the question of
the desired number of classes - the example uses 5, but it could have been
something else.

If you have "ground truth" - some classified data - you can fit a
classification tree (rpart() function in the standard package of the same
name), but without a "true" starting classification, it isn't easy.

So you'll need to read up on classification with categorical variables, 
I'm afraid.

Roger

> 
> Martin
> 
> On Thursday 01 March 2007 09:35, Andruit at gmx.de wrote:
> > Hi,
> > I`m a newbie in R and haven´t used it many times....
> >
> > I have several Rastermaps of the same area (vegetation, geology, wetness,
> > aspect, slope ....) Now i would like to find areas having similar
> > characteristics...
> > How can I do it???
> > I was thinking about cluster analysis in R.
> > Has anybody done it before? I think so.
> >
> > Andi
> 
> _______________________________________________
> statsgrass mailing list
> statsgrass at grass.itc.it
> http://grass.itc.it/mailman/listinfo/statsgrass
> 

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no