/higgsml-sklearn

A demonstration of scikit-learn using the Higgs Boson Machine Learning Challenge dataset.

Primary LanguageJupyter Notebook

higgsml-sklearn

A demonstration of scikit-learn using the Higgs Boson Machine Learning Challenge dataset.

  1. Download the data file:

    python download_data.py

  2. Start up the Jupyter notebook server:

    jupyter notebook

  3. Click on Tutorial.ipynb in the browser to walk through the tutorial.

Scripted analysis

A set of scripts for training and analysis are in the python directory. From the project directory, run python python/BDT200/train.py to train the TMVA-like model. Then run python python/BDT200/analysis.py to calculate the expected discovery significance.

Tips & tricks

Working over SSH

You may wish to run on a more powerful, air-conditioned machine over an SSH connection. In this case, use port forwarding to access the Jupyter notebooks. In Step 2 above, start the notebook server with

jupyter notebook --no-browser

There will be a line similar to

[I 17:22:38.962 NotebookApp] The Jupyter Notebook is running at: http://localhost:XXXX/

in the output, where XXXX is the port number (default is 8888). Use the SSH magic key combination, normally Enter+~+C to enter the SSH prompt. Then enter -L XXXX:localhost:XXXX and point the web browser on your local machine to http://localhost:XXXX/.