Preprocessing issue in Data Analysis of ttbar using DeepCSV

ROOT
1

Hello Everyone,
I’m trying to learn the machine learning tool to discriminate b-tagging among c and light quark jets.
In the pre-processing, when I’m doing the following step:

  1. Padding the missing value to zero.
  2. Scaling the variable and then adding the offset to each feature.
  3. I have also tried input_value = input_value - feature.mean().

All the information are collected from the github: RecoBTag-Combined/DeepCSV_PhaseI.json at master · cms-data/RecoBTag-Combined · GitHub

Data is taken from the latest:
/eos/cms/store/group/phys_btag/ParT_2024/ttbar_test

But their is discrepancy between the model predicted and actual jet tag.

Let me know if I’m missing anything in this.

2

Hello,

Thanks for the post.
This looks more like a question for the CMSTalk forum: have you tried to post this on that platform?

Cheers,
D

3

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.