[R-sig-Geo] gstat model cross validation SSerr vs other diagnostic measures

[Edzer Pebesma]

They are very different things. SSErr is an attribute of a variogram
model fitted to a sample variogram that describes how well this fit
went. Cross validation statistics tell you how well prediction went in a
cross validation context, under a given variogram model.

You can for instance multiply the variances in your variogram model with
a constang, e.g. change vgm(1, "Exp", 3, 1) with vgm(2, "Exp", 3, 2) and
you'd see completely identical predictions, but SSErr values typically
be very different.

On 04/02/2014 11:37 PM, Moshood Agba Bakare wrote:
> Dear All,
> Could any one tell me the difference between SSerr (Sum of Squares error)
> differs from other gstat krige.cv cross validation diagnostic measures such
> as mean error (ME), root of means square error (RMSE) and means square
> deviation error (MSDR) of residual to kriging variance?
> 
> is SSerr appropriate statistics for validating a model? Is SSerr means sum
> of squares of difference between actual values and predicted values of
> fitted model? in MSDR, what is difference between residual and kriging
> variance? formula for computing the kriging variance.
> 
> Thanks
> Moshood
> 

-- 
Edzer Pebesma
Institute for Geoinformatics (ifgi), University of Münster
Heisenbergstraße 2, 48149 Münster, Germany. Phone: +49 251
83 33081 http://ifgi.uni-muenster.de GPG key ID 0xAC227795

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 490 bytes
Desc: OpenPGP digital signature
URL: <https://stat.ethz.ch/pipermail/r-sig-geo/attachments/20140403/50c110ee/attachment.bin>