On May 24, 2006, at 12:41 PM, Alberto Cuesta wrote:
> Hi again, I have usually to test the individuals I create with ECJ.
> The usual way, you train a population with one set of data, save
> your best individuals to disk, and then you have to test them with
> other different set of data.
> The thing is, I'm not really sure how this has to be done, I've
> rewritten ec.Evolve to do this, but it's a big file and I don't
> fully understand what I'm doing.
> Can I find somewhere a file with this procedure? the most basic
> thing, from a parameter file and a individual stored in another
> file, get to run the evaluate method from the problem stated in the
> parameters file.
I'm not sure why this needs to involve Evolve at all.
If you are trying to evaluate a population of individuals stored in a
file, you should be able to add a parameter collection like this:
pop.subpop.0.file = albertos/file/of/individuals.to.input
generations = 1
You'll need to either specify in the file that the individuals
haven't already been evaluated, or set up your problem to ignore the
evaluated flag and reevaluate them anyway.