Usage: Compute mean/std/min/max of model errors (good, bad, mean) and could make an histogram.

Options:
  -h, --help            show this help message and exit
  -f FILTER, --labelsfilter=FILTER
                        only models matching this filter are covered.ex :
                        'label1,label2' or 'label' or 'regexp1,regexp2' (here
                        order is important)
  --filtermode=FILTER   strict, flexible or regexp
  --filter-fifty-percent
                        filter values equal to 0.5
  -m MODEL, --model=MODEL
                        model file name (default : model.arg)
  --hist=HIST           make histogram from classifiation/regression ratesof
                        models. HIST should be a format supported by pylab