Zero nugget, but still residuals in cross-validation ?

[LineThomsen]

In my present project I am using ordinary kriging and comparing various models. I let the program decide on the nugget, it is set to TRUE and the measurement error model is set to 100%. Then it happens that the program comes up with a kriging-model where the nugget says zero, fx: Model : 0*Nugget+19742*Stable(154.68,1.3197)

but when I further in the wizard arrives at the Cross-validation, there is still prediction errors present. As I understood it, when the chosen model has a nugget of zero, the model will be an exact interpolator and so there should be no difference between the measured values and the predicted? Can someone please help me understanding this?

Best regards

L. Thomsen

[EricKrause]

With no nugget, kriging is an exact interpolator.  If you use GA Layer to Points and predict back to the input point locations, you should get perfect predictions.  Specifically, the input point will get a weight of 1 and all other neighbors will get a weight of 0.

In crossvalidation, you are throwing out the input point before predicting back to that input point location, so it can't just assign it a weight of 1 (because it has been thrown out of the dataset).  That is why an exact interpolator will still have crossvalidation errors.

[LineThomsen]

ok, thanks for the answer - it was useful 🙂