Well, after the previous post, I did write a (or part of a) data file, and ran that through WEKA. The results given didn’t help at all (likely due to the data representation) so I just ran another proving run with the last working agent and after that, ran my testing run.
Unless one of the top 3 don’t run their testing or somehow my agent can cope with the ‘hidden surprise’ well, I’ll come in 4th (or maybe worse. I hope not).
Just waiting on the results and the released source code now.