"""
tsne.py
This script produces an interactive t-SNE plot of the events inside an HDF5 file. No changes to the file are
required, only command line argument. Run the file with the -h flag, to get a description of all available arguments.
"""
# ---------------------------------------------
# Imports
# ---------------------------------------------
import cait as ai
import os.path
from os import path
import argparse
import numpy as np
import matplotlib as mpl
# ---------------------------------------------
# Read command line arguments
# ---------------------------------------------
# Construct the argument parser
ap = argparse.ArgumentParser()
# Add the arguments to the parser
ap.add_argument("-f", "--file_name", type=str, required=True, help="The name of the HDF5 file.")
ap.add_argument("-c", "--channel", type=int, required=True, help="The channel number within the HDF5 file.")
ap.add_argument("-t", "--time_series", action='store_true', help="Use the raw time series as data instead of the main parameters.")
ap.add_argument("-a", "--additional_mainpar", action='store_true', help="Use the additional main parameters as data instead of the main parameters.")
ap.add_argument("-s", "--test_size", type=float, required=False, default=0.5, help="The share of events used as test set.")
ap.add_argument("-i", "--highest_idx", type=int, required=False, default=None, help="The highest index that is included in the plot. Use this if there are too many events to handle for the scatter plot.")
ap.add_argument("-p", "--perplexity", type=int, required=False, default=30, help="The perplexity value, hyperparameter of the t-SNE plot. Recommended value: Between 5 and 50.")
ap.add_argument("-m", "--matplotlib", action='store_true', help="Start matplotlib for plotting, instead plotly.")
args = vars(ap.parse_args())
assert args['time_series'] is False or args['additional_mainpar'] is False, "You cannot activate both time_series and additional_mainpar!"
assert args['test_size'] > 0 and args['test_size'] < 1, "The test_size must be between 0 and 1!"
assert path.exists(args['file_name']), f"The file {args['file_name']} does not exist!"
assert args['channel'] >= 0, 'You cannot choose a negative channel number!'
if args['highest_idx'] is not None:
assert args['highest_idx'] >= 1, 'The highest index must be 1 or higher!'
assert args['perplexity'] > 1, 'The perplexipy must at least be 1. Recommended value: Between 5 and 50.'
# ---------------------------------------------
# Main
# ---------------------------------------------
if args['time_series']:
which_data = 'time_series'
elif args['additional_mainpar']:
which_data = 'additional_mainpar'
else:
which_data = 'mainpar'
if args['highest_idx'] is not None:
only_idx = np.arange(args['highest_idx'])
else:
only_idx = None
et = ai.EvaluationTools()
et.add_events_from_file(file=args['file_name'],
channel=args['channel'],
which_data=which_data,
only_idx=only_idx,
)
et.split_test_train(test_size=float(args['test_size']))
if args['matplotlib']:
et.plt_pred_with_tsne(pred_methods=[],
what='all',
verb=True,
perplexity=args['perplexity'],
)
else:
et.plt_pred_with_tsne_plotly(pred_methods=[],
what='all',
verb=True,
perplexity=args['perplexity'],
)