Raw and weighted yields in significances output is incorrectly multiplied by luminosity

[jmrolsson]

https://github.com/kratsg/optimization/blame/master/root_optimize/command_line.py#L177

This should be fixed so that it's only scaled if the count_type is scaled.

[kratsg]

Refers to this line:

optimization/root_optimize/command_line.py

Line 177 in 730616f

sig_dict = dict([('hash', cuthash)] + [('significance_{0:s}'.format(counts_type), utils.get_significance(args.lumi*1000*counts, args.lumi*1000*total_bkgd[cuthash][counts_type], args.insignificanceThreshold, args.bkgdUncertainty, args.bkgdStatUncertainty, total_bkgd[cuthash]['raw'])) for counts_type, counts in counts_dict.iteritems()] + [('yield_{0:s}'.format(counts_type), {'sig': args.lumi*1000*counts, 'bkg': args.lumi*1000*total_bkgd[cuthash][counts_type]}) for counts_type, counts in counts_dict.iteritems()])

This is going to be improved by removing lumi calculations entirely. The suggestion will be that the correct weight is applied, instead of having an arbitrary scaling passed in through the CLI -- and instead by creating a branch (or an alias) for that arbitrary scaling as needed.