PLAN for cocomin experiment:

compete script that calls with each data file like this:
./cocomin_experiment $Path $File

cocomin experiment is c
	does an n-fold experiment with
	RankerSet="none correlation variance_high variance_low entropy_high entropy_low"
	EvalSet="mmre sd_mre pred30 correlation"
	HorizonSet="0 1 2 4 8 16"
	SearchSet="forward backward both(separately)"

	N * 6 * 4 * 6 * (3 the search set isn't really more iterations) = 432 N

monstrous task of analyzing results when the file is that big.

after the experiment is done, first show Horizon 16 as superior and prune results files:
  now it's only 27 N


cmd | tee -a outfile