generate_features.py

########################################################################
#
# @author : Emmanouil Sylligardos
# @when : Winter Semester 2022/2023
# @where : LIPADE internship Paris
# @title : MSAD (Model Selection Anomaly Detection)
# @component: root
# @file : generate_features
#
########################################################################

import numpy as np
import pandas as pd
import argparse
import re
import os

from utils.data_loader import DataLoader

from sktime.transformations.panel.tsfresh import TSFreshFeatureExtractor


def generate_features(path):
	"""Given a dataset it computes the TSFresh automatically extracted 
	features and saves the new dataset (which does not anymore contain
	time series but tabular data) into one .csv in the folder of the
	original dataset

	:param path: path to the dataset to be converted
	"""
	window_size = int(re.search(r'\d+', path).group())

	# Create name of new dataset
	dataset_name = [x for x in path.split('/') if str(window_size) in x][0]
	new_name = f"TSFRESH_{dataset_name}.csv"

	# Load datasets
	dataloader = DataLoader(path)
	datasets = dataloader.get_dataset_names()
	df = dataloader.load_df(datasets)
	
	# Divide df
	labels = df.pop("label")
	x = df.to_numpy()[:, np.newaxis]
	index = df.index

	# Setup the TSFresh feature extractor (too costly to use any other parameter)
	fe = TSFreshFeatureExtractor(
		default_fc_parameters="minimal", 
		show_warnings=False, 
		n_jobs=-1
	)
	
	# Compute features
	X_transformed = fe.fit_transform(x)

	# Create new dataframe
	X_transformed.index = index
	X_transformed = pd.merge(labels, X_transformed, left_index=True, right_index=True)
	
	# Save new features
	X_transformed.to_csv(os.path.join(path, new_name))


if __name__ == "__main__":
	parser = argparse.ArgumentParser(
		prog='generate_features',
		description='Transform a dataset of time series (of equal length) to tabular data\
		with TSFresh'
	)
	parser.add_argument('-p', '--path', type=str, help='path to the dataset to use')
	
	args = parser.parse_args()
	generate_features(
		path=args.path, 
	)