[R-sig-Geo] R2 values from SSErr fit.variogram attribute

[Edzer Pebesma]

On 03/15/2011 04:18 AM, Sébastien Durand wrote:
> Hello Dr. Pebesma and all other 1989 readers...
> 
> Believe me, I would have been much happier, if I could comprehend this tiny little detail that you refer to as : 
> 
>> It looks quite OK; I suspect however that instead of using
>> mean(s.v$gamma) for SSTot, for this mean the same weighting scheme
>> should have been used.
> 
> I still do not understand what you mean by that.  I am very sorry for that and yes I now feel stupid, sincerely I am doing my best to graps very old notions.
> 
> In my function, it is the weighted residual that are squared, that part is ok, but now you are telling me that I am not using to proper "mean" reference, that in fact the "same weighting scheme should be used on that ... ???  Did you meant "mean(vario$gamma)*weig" ??? 
> 
> I would like to receive an explanation for this?

You're comparing a weighted fit to an unweighted mean. If you want to
evaluate the fit (by R2), the corresponding null-model should use the
same weights, IMO. Given weights w and values sv$gamma, that would be
sum(w * sv$gamma)/sum(w) rather than mean(sv$gamma).

Or, alternatively, weighted.mean(sv$gamma, w)

> Sincerely!
> 
> 
> The other software is GS+
> 
> S.
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo

-- 
Edzer Pebesma
Institute for Geoinformatics (ifgi), University of Münster
Weseler Straße 253, 48151 Münster, Germany. Phone: +49 251
8333081, Fax: +49 251 8339763  http://ifgi.uni-muenster.de
http://www.52north.org/geostatistics      e.pebesma at wwu.de