here my problem.
I am creating Histograms for different MC dijet samples (J0-J6) individually. The problem now with creating the Efficiency is that obviously in the different samples the statistics is different to a point where there are no entries. I have a feeling that J0 and J1 samples (very high weights but no statistics) contribute as huge errors to my efficiency curves.
If I only for example take the J3 sample I have errors of about 2 % if I add the J0 and J1 (basically no entries in the interesting region) my errors are about 40%.
I am using the “mode” option. Without it not only the errors are going to 50% but also the whole efficiency curve is at pushed to 0.5.
In addition I was wondering if there is a way to not fill the bins where none of the samples has an entry because as it is right now they are filled with a value about .5 and an error of .3
thanks for the help
so I think I know where the problem is. I understand that there is a norm factor (basically takes into account that one entry with a weight 1000 needs a huge error)(norm = sum(weights)/sum(weights^2)). What I did now is that I took out the weight sum contributing to the total norm in case an entry (denominator) is zero.
I am not sure if that is absolutely correct but otherwise the errors are overestimated I believe