Gerhard Fürnkranz wrote:
my experience is that the verification result depends significantly on the distribution of
> the test set points used to assess the "resulting profile accuracy". My feeling is that I get
in a majority of cases better average errors, whenever the same kind of distribution is usedfor both, the training set and the test set.
Yes, this is my experience too. That's why in testing this, I ended up using a very large number of purely randomly distributed test points during verification. Graeme Gill.