Generate datasets

Note : Run all the scripts from the root directory adn_circulant

First check all the data (metadata data and raw data) are present in the directory data/raw

Then run the following command to format data and merged raw data with metadata; it will create 3 files:

• merged/diag_data.csv
• merged/chir_data.csv
• merged/end_data.csv

python data_processing/load.py

Then you need to generate features and labels using the script compute_features.py.

This script takes 3 arguments (they all have default values), and you can find information about these parameters in the code or by executing the command :

python data_processing/compute_features.py --help

Run multiprocessor training

You can run the training of many random models by using the script multiprocessing_training.py. This script has parameters that you can find about using the command :

python multiprocessing_training.py --help

It will log all the results in a file name adn_circulant_brute_force_fold_i.log, where i corresponds to the fold the training is done on.