 |
 |
 |
 |
| [NN |
[Internal Page] |
Neural Network b-Tagger for SecVtx-Tagged Jets
Comparing Data and Monte Carlo
Neural network outputs for the inclusive electron sample:
|
|
|
|
Electron jets are are very pure in b jet composition; Away jets are likewise, but to a slightly lower extent. As we see, the agreement between data and Monte Carlo is very good. input distributions may be viewed here.
Using the network cuts that produce 90% tt b efficiency, we obtain the efficiencies and scale factors in the table below. Since the electron jets are more pure in b than away jets, 0.97 is taken to be the b efficiency scale factor. The uncertainty comes from the difference between electron jet SF and away jet SF, doubled: 0.97 ± 0.02. The scale factor is confirmed via paramterization by jet ET.
| Data Eff. | MC Eff. | Scale Factor | |
|---|---|---|---|
| Electron Jet | 0.874 ± 0.005 | 0.898 ± 0.006 | 0.973 ± 0.009 |
| Away Jet | 0.855 ± 0.005 | 0.889 ± 0.006 | 0.964 ± 0.009 |