[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Ivo Thonon wrote:
> Rebecca wrote:
> I might be really daft but is there an automatic way of getting the
> goodness-of-fit statistics for fitted variograms?
AFAIK, there is no `the' goodness-of-fit statistic: several
statistics can measure goodness-of-fit. When you start gstat as
gstat -d65 -lgstat.log whatever.cmd
it will write fit diagnostics to the logfile gstat.log.
The only diagnostic printed is the (weighted) sum of
squared errors of the fit, for each iteration. This is
the criterion that is minimized in the fit. Note that
it (strongly) depends upon the weights you choose (fit
What Ivo referred to is R2. As R2 measures the performance
of a fit against the fit of a reduced model, it fails when
there is not a suitable reduced model, which is often the
case when modelling variograms (e.g. when no nugget is fit).
Hope this helps,