Cross-validation

[Omodolor@1]

We are having this issue:
flag.est and flag.std both produce two variables. The first variable in flag.est makes sense, it is either the estimation error (esterr) or the estimate (est) depending on whether or not the use a 1 or -1. The second variable is questionable... It is called stderr, but the numbers are not correct for the standard error of the estimate. For flag.std, the first variable created is the esterr (estimation error) which seems odd as the estimation error (if not standardized) is Z*-Z - I get that in the first flag.est! either (Z*-Z)/S or S. The second variable created is either stdev (which seems to make sense) or stderr which could make sense, but the numbers are identical that of sterrr in flag.est, and make absolutely no sense. I would expect to have the standard dev (std) and standardized error std/(sqrt n) or something like that...

Can they provide us with a better explanation?

[Didier Renard]

Hello
I have checked that there was still a bug hanging there in the Cross-validation procedure.

I corrected the underneath library but I am not sure when this patch will be available in the next version of RGeostats.

In the mean time, I can simply encourage you to run the xvalid.Rmd script (available in the Documentation tag of this website) and check that the results, as they are explained, are the ones that you obtain with your version.