Joint fit of binned distributions

Dear experts,

I’d like to perform a joint fit of two distributions with a simPdf. The problem is I only have the binned distributions of the data: is it possible to perform the fit on a RooDataHist instead of a RooDataSet? Is there some example of a fit of this kind?
Sorry but I cannot find any hint on the manuals.



Unfortunately, I don’t know if that’s possible! Maybe I can just suggest a workaround if you’re interested for the time being: to transform your 100 bins histogram into 100 entries with a weight variable corresponding to the value in that bin. Then you can add a category variable for each of the two histograms (now dataset).


– Gregory

Hello Gregory,

thanks for your suggestion, I’m trying do develop it into some working code.
In the meanwhile I tried to pick up and fit the histograms in the following way:

[code]RooDataHist binnedData(“binnedData”, “joint data”,
w::mass, Index(efficiencyCategory),
Import(“Passed”, *hist_passing),
Import(“Failed”, *hist_failing) ;


where hist_failing and hist_passing are the TH1 histograms, efficiencyCategory is the RooCategory and simPdf is the RooSimultaneous pdf. I don’t know if that is properly recommended, but the fit looks ok!
By the way I have some troubles with the number of events in the RooDataHist, wich is doubled with respect to the original TH1 histogram: this is very strange, since the default weight applied to the TH1 by the Import class is 1.


Can I ask if there was any solution / workaround to this problem?
I think I’m experiencing the same problems.

Thanks in advance,