Missing weight in normalization denominator about klubanalysis HOT 6 CLOSED

Yes exactly, the denominator must contain only the weights that changes the shape and not the normalization.

from klubanalysis.

I think I would agree.
(I was looking at the same part of the code for the b-tag SF normalization...)

from klubanalysis.

Hi Marcel! I am not sure about the b-tag scale factors. IIRC the dedicated procedure that needs to be performed should preserve the normalization.

from klubanalysis.

Yes, indeed it will be somewhat different for the b-tag SFs where we actually need two of these terms you linked above (for a specific selection with and without b-tag cuts).

So to be 100% sure, the terms that are multiplied to EvtW above are meant to come from particular weights that are not intended to change the overall rate, right?

from klubanalysis.

I created #305 to include this fix early on, since we could use the weekend to rerun our small ntuples :) (of course given that the discussion converges)

from klubanalysis.

Closing the issue, since it appears not to be one in the end

from klubanalysis.