Poisson Goodness of Fit

[Wrong Sign D Group] [CLEO] [HN Home Page]

A CBX note on Goodness of Fit

Often, contemporary HEP analyses seem to omit goodness of fit studies. This is unusual, because a result with a poor goodness of fit should be interpreted with caution.

For binned fits, where every bin has more than 20 or so events, one should be able to use the chi-squared technique.

For totally unbinned fits, Tim Nelson and I wrote CBX 98-61, available here, which contains a description of a goodness of fit test for the unbinned case.

For the intermediate case, pertinent for the current D0 > K+ pi- `wrong sign' analysis, I have written the following CBX. By the way, a key part of the technique used for the wrong sign is the Poisson subtraction technique developed by Jeff Gronberg for his D*pi analysis, and described in an Appendix to CBX 96-69a, available here.

So, the postcript of the current CBX draft is here, and the raw TeX file is here.

The postscript of the figures, and the text of the PAW kumac is available here:

Figure 1: ps , eps
Figure 2: ps , eps
Figure 3: ps , eps
Figure 4: ps , eps
PAW kumac fin.kumac

[Wrong Sign D Group] [CLEO] [HN Home Page]