I am using the data driven method with RooFit to fit the Sidebands data to have the fitting function, then with this function to obtain the expected background in the Signal Region.
And I have the normalize RooFit function as the pdf, so I have the integral probability of the Sideband and the Signal Region. As I know the total number of events of Sideband, then through the probability of Sideband and the Signal Region obtained from the pdf function, I can find the fitted events in the Signal Region.
And when I use the “getPropagatedError” to get the error of the Signal Region, I found the propagated error of signal region is equal to the error of the sideband. So the error of overall Region(Signal Region+Sideband) is 0, which I think it’s weird because I presume that as I treat the number of events of Sideband to be fixed, the error of sideband should be 0, rather than the whole Region(Signal Region+Sideband).
Actually the results is
n_Sideband= 1234±40.1432
n_SignalRegion= 266.297±40.1432
n_WholeRegion= 1500.3±1.66566e-13
Does anyone know why this happen?
Thank you so much!
Hai