Skip to content

Latest commit

 

History

History
27 lines (19 loc) · 1.55 KB

FITTING.md

File metadata and controls

27 lines (19 loc) · 1.55 KB
Raw

Statistical analysis

This readme describes how to perform the statistical analysis on the histograms produced by ewkcoffea.

Making datacards

The information in the histograms (produced by the processor) can be converted into datacards (to be fed to combine) via the make_datacards.py script.

  • Example of running over Run2 histograms with systematics (-s) for the BDT bins (--bdt). 
    python make_datacards.py histos/yourhist.pkl.gz --run run2 -s --bdt

Note that by default the code will include CR bins to handle background normalization via a simultaneous SR+CR fit, using the rateParam functionality in combine (described in the combine documentation here).

Using combine to perform the statistical analysis

The statistical analysis is performed with the combine tool. This should be set up outside of your conda env, following the instructions here.

Copy the datacards (from ewkcoffea) to your combine working area. To combine the cards from all of the channels into a single card, use the combineCards.py command, e.g. run the following (in a directory where wwz4l_card_* match to all of your cards and no extra cards):

combineCards.py wwz4l_card_* > comb.txt

To check the significance, run the following:

combine -M Significance comb.txt --expectSignal=1 -t -1