Perhaps the simplest approach would be to override the
SimpleProblemForm.describe(...) method, which is called by
KozaStatistics at the end of the run.
On Oct 2, 2006, at 6:19 AM, John Woodward wrote:
> I am using ECJ for Genetic Programming.
> Once I have evolved an individual, I would like to test its
> performance on an independent set of test cases to see how it
> generalizes. How do I do this?
> The best individual is printed out in KozaStatistics.finalStatistics
> It can be accessed as
> However, I want to evaluate this individial on a seperate set of
> test cases.
> I could write a seperate method (e.g. generalization) in my problem
> class, but how would I call it?
> The generalization method will be identical to the evalute method in
> my problem class, but on a seperate set of data.
> Any ideas?
> I am sure someone must have done this before if they have coded GP
> John Woodward.