[R-sig-Geo] 1. Re: default value for cutoff in gstat variogram()

[Paulo Justiniano Ribeiro Jr]

Hi
My 2 cents here.

I agree with Edzer that from statistical point of view there is not much 
point in compare this.
However the software implementation and checking can benefit from it.

Let me try to remember a few possible relevant points for the diferences:

1. the values in the x-axis : some will average distance values while 
others will take the mid point of the classes intervals (as done, for 
instence in histograms)

2. How to deal with interval borders (<= ou <, >= or >= )

3. How to deal with coincident locations

4. How the maximim distance for the variogram is defined

5. How to deal with NA's in some distance classes and minimum number of 
pairs

The packages will have different default options (maybe not aways clearly 
documented).
A commom exemple run using all packages would be welcome, and this should 
include not only geoR and gstat but also the opthers.
We only have to bear in mind that the choice of empirical variograms will 
aways be subjective, according to the options defined by the user.

Best
P.J.





On Wed, 7 Jun 2006, Edzer J. Pebesma wrote:

> Date: Wed, 07 Jun 2006 08:59:00 +0200
> From: Edzer J. Pebesma <e.pebesma at geo.uu.nl>
> To: john lewis <john.lewis at sympatico.ca>
> Cc: r-sig-geo at stat.math.ethz.ch
> Subject: Re: [R-sig-Geo] 1. Re: default value for cutoff in gstat variogram()
> 
> john lewis wrote:
>> Hello all,
>> I did a comparison of gstat, geoR and SPlus 6.1 about 2 years ago with the coal ash data and found little difference among them after a bit of time making sure all input parameters were the same. I don't know if I can find those results but maybe it is time for me to do it again?
>>
> Maybe they're not worrying from a statistical point of view, but from
> the point of view of code verification I am interested to see them (with
> the data & scripts to generated them). The coalash data are only
> available in S-Plus, am I right? (We do have an S+SpatialStats licence,
> but never got around to set it up for 7.0)
> --
> Edzer
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>

Paulo Justiniano Ribeiro Jr
LEG (Laboratório de Estatística e Geoinformação)
Departamento de Estatística
Universidade Federal do Paraná
Caixa Postal 19.081
CEP 81.531-990
Curitiba, PR  -  Brasil
Tel: (+55) 41 3361 3573
Fax: (+55) 41 3361 3141
e-mail: paulojus at est.ufpr.br
http://www.est.ufpr.br/~paulojus