# About Roofit use manual's Appendix C

Hello,

In RooFit_Users_manual_2.07-29, I can not find Appendix C. where can I find it?
If I want to know the differences between the default fitting method in ROOT and in RooFit, whether I can get the answer from Appendix C?

Thanks
foolfish

Hello,

The authors wrote in the Appendix C of the manual “This section is scheduled for the next version of the manual” (see page 69). I have pointed your post to Wouter Verkerke and I hope he will answer your question.

Cheers, Ilka

Hi,

foolfish

Hi,

Appending C is still work in progress (unfortunately).
Is there a specific question I can answer for you meanwhile?

Wouter

Hi, Wouter

I am sorry to reply later . I have two question:

1. Chi-squre fit method and Loglikelihood fit method, what are their advantage and disadvantage? If I want to fit a TGraph, which one should be chosen?
2. In ROOT, if I fit a TGraphErrors with Chi-squre fit method, the weight how to be computer?

Best regard!
foollish

Hi,
sorry for the late replay.

See for example section 32.1.2 and 32.1.3 in

pdg.lbl.gov/2004/reviews/statrpp.pdf

For a TGraph in general you should use the chi2 method and this is what is done in ROOT when fitting via TGraph::Fit.
In some special cases, for example when fitting efficiencies, which have the binomial errors, then it is recommended to use a log-likelihood method.

The weights used when fitting a TGraphErrors takes into account also the error on the x value (coordinate).
The effective variance method is used. The chi2 element are computed as
the sum of the quantity below at each point:

``````                 (y - f(x))**2
-----------------------------------
ey**2 + (ex*f'(x))**2
``````

where ey is the error on the value y, ex is the error in x.

Best Regards

Lorenzo