bdrum / cern-physics

My analysis workflow
1 stars 1 forks source link

Unbinned fitting (Likelihood) #50

Open bdrum opened 3 years ago

bdrum commented 3 years ago

For fitting in #44 I have standard $\chi^2$ test for deterring quality of the fit, but for area with small statistics (Puasson) it's wrong. E.g. I've calculated error as $\sqrt(N)$ for each bin, but in case of small count of events in bin it provide us the situation when 2 events in bin and 1.4 sigma.

Here is the fragment from lection of Cambridge prof. Mark Thomson about the question:

image