spec.json
taken form @kratsg's example Gist.
bash run.sh
# pyhf cls spec.json
{
"CLs_exp": [
0.3779766122079999,
0.5257773520036446,
0.702130281843784,
0.8695189282227754,
0.9691693123561054
],
"CLs_obs": 0.5932002295253658
}
# python run_cls.py HistFactory/config/FitConfig_combined_meas_model.root
# pyhf xml2json --basedir HistFactory --output-file roundtrip.json HistFactory/FitConfig.xml
# pyhf cls roundtrip.json
{
"CLs_exp": [
0.37796298099083625,
0.5257645108789305,
0.7021204342074518,
0.8695137249328984,
0.9691678471392434
],
"CLs_obs": 0.5931943069940612
}
Looking visually at scan.pdf
the observed CLs value at mu=1
seems to be below 0.6