Fit my model to data in binned Poisson Likelihood

Dear experts,

I am trying to fit binned data with simulated background+signal. I have an analytical expression for the signal in each bin, which is a quadratic order polynomial of two parameters. Due to low statistics, I need to use Poisson likelihood and not gaussian pdf. I need to get the 95% C.L. of parameters using Profile Likelihood Ratio. I am new to Roofit and RooStats. Could you help how to go about it? I was going through the example macro given in ROOT: tutorials/roostats/rs_numberCountingCombination.C File Reference . However, I am not able to understand how to give the binned analytical expression for the signal events that I have. Could you please help?

Best regards,

I think @jonas can help you

A response to the query would be helpful…


May be @moneta also knows.

If you want to fit your binned data with a model and then estimate the 95% confidence intervals of the parameter using the profile likelihood ratio, you would need to perform a binned likelihood fit.
You organize first your data as a RooDataHist object and the you can model the signal function using standard RooFit classes. An example is the basic tutorial rf102_dataimport.C.
Afterwards you can either estimate the confidence intervals directly using Minos, or use the RooStats ProfileLikelihoodCalculator, as shown in this RooStats tutorial. If you have questions, please let me know