Fits with Unsymmetrical error

Roofit and RooStats
1

Hi
can any one guide us that how to fit the function and minimize chisq when one variable in data has unsymmetrical error?

i.e.
I want to fit a two parameter fit function (+1 normalization) to data in x and y. There is no error in x but y has unsymmetrical error.
i.e y+(dy1)(-dy2) where dy1 and dy2 are unsymmetrical errors.

2

Hi,

You can store your data (x,y) in a TGraphAsymmErrors and use the TGraphAsymmErrors::Fit function, that does a least square fit to the data by using the upper error when the function value is larger than the y point and the lower error in the opposite case Function value lower than y value)

Best Regards

Lorenzo

1 Like
3

Thanks a lot Lorenzo. I will implement.
manjit

4

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.