Friday, February 25, 2011

Restarted HCV linkage

Things were taking too long on the HCV linkage and the p-values were "inconclusive". I changed the code so it won't calculate p-vals unless asked (otherwise it will just leave that column blank). I restarted the code on sha a792bca5d5d14506b83f1af96a0f0c3201a5c232 and command:
python AnalysisCode.py --link-lanl --data-dir HCVData/ --workers 3 --max-width 5

I'll create a function which can go back and recalculates the p-values.  I also need to deal with "inconclusive" p-values ... which I'm defining as permutation tests where I never (or only twice) find a value as good as the real value. In a set of 5000 iterations I don't have a good estimate.  So I think I'll refactor that function to keep testing until it finds 10 "good" values unless it gets to 100,000 iterations.  I'll probably have a minimum of 5000 just to make sure it doesn't get a "lucky break" and find those 10 early.

No comments:

Post a Comment