Analysis of PINPOINT data

• preprocessing.py modifies data in data directory:
  • By default this should be outside of the analysis directory, path can be changed in get_data_path function in utils.py
  • Expects common ROOT files in data/root directory and will create new parquet and torch directories with preprocessed data
• cnn.py runs neutrino flavor identification using 2d CNNs with zx and zy projections of binned pixel hits. This is meant to be run on a remote machine with a powerful GPU
• evaluate_cnn.ipynb takes weights from cnn.py and is meant to be run locally to produce performance plots