CMS-ML-analysis

A repository to aggregate some tools for HEP analysis using NanoAOD format, also using ML. We are following some of the contents from the last CMS Open Data School [https://cms-opendata-workshop.github.io/2024-07-29-CERN/]

Besides the usual python packages, to do a modern analysis, we will need some additional tools.

The most important ones are:

• awkward
• uproot
• xrootd
• vector
• coffea

Requirements

For these analysis, I used:

• working PC
• internet
• Ubuntu 24.04
• python3.10

Preparation

Install packages

First, if you are not using a docker, install those:

$ sudo apt-get install cmake uuid-dev libssl-dev xrootd-client xrootd-server python3-xrootd

Create a virtual enviroment

python -m venv .venv

You can also choose another name, like .cmsml, or .myvenv

After the .venv creation

This could take a while

$ pip install -r requirements.txt $ pip install --upgrade pip $ pip install --upgrade awkward $ pip install --upgrade uproot $ pip install --upgrade pandas $ pip install --upgrade coffea $ pip install --upgrade matplotlib

Next step: Follow the notebooks

In the notebooks folder. For this, I recommend using VSCode with Jupyter extensions (my case) or Jupyterlab itself.

TODO:

• Create a first principles notebook
• Update akward arrays and uproot to coffea [?]
• Configure and make functions plots.
• Introduce some part of ML prject.
• Wrapup the main functions in streamlit
• Go live.