signalytica 0.1.17

Feature extraction for time series and EEG signals

SignaLytica

A Python package for feature extraction in time series and EEG signals

Installation

pip install signalytica

Dependencies

• NumPy
• SciPy

Utilization

Import modules

import numpy as np
import matplotlib.pyplot as plt
from signalytica import SignaLytica

Example time series data

np.random.seed(1234)
x = np.random.normal(size=128)
plt.figure(figsize=(6,2))
plt.plot(x)
plt.title('Example time serie'); plt.xlabel('Samples'); plt.ylabel('Amplitude')
plt.tight_layout()
plt.savefig('example_time_serie.png')

Output:

List all the available metrics in the package

signa = SignaLytica()
available_metrics = signa.list_metrics()
print(available_metrics)

Output:

['mean', 'std', 'coeff_var', 'median', 'mode', 'max', 'min', 'first_quartile', 'third_quartile', 'inter_quartile_range', 'kurtosis', 'skewness', 'activity_hjorth_param', 'mobility_hjorth_param', 'complexity_hjorth_param', 'total_power_spectral_density', 'centroid_power_spectral_density', 'relative_delta_power', 'relative_theta_power', 'relative_alpha_power', 'relative_beta_power', 'relative_gamma_power', 'determinism', 'trapping_time', 'diagonal_line_entropy', 'average_diagonal_line_length', 'compute_recurrence_rate', 'spectral_edge_frequency_25', 'spectral_edge_frequency_50', 'spectral_edge_frequency_75', 'delta_amplitude', 'theta_amplitude', 'beta_amplitude', 'alpha_amplitude', 'gamma_amplitude', 'hurst_exponent']

Extract specific features from time serie/EEG signal

features = ['coeff_var', 'inter_quartile_range', 'kurtosis', 'total_power_spectral_density', 'trapping_time']
signa.extract_features(x, features)

Output:

{ 'coeff_var': 0.9296511098623816, 'inter_quartile_range': 1.1642928949004765, 'kurtosis': 0.9355744131054928, 'total_power_spectral_density': 1.03503685376608, 'trapping_time': 2.2205882352941178 }

Calculate individual features

Activity Hjorth Paramater

hurst_exp = signa.activity_hjorth_param(x)
print(hurst_exp)

Output:

0.9296511098623816

Total power spectral density

hurst_exp = signa.total_power_spectral_density(x)
print(hurst_exp)

Output:

1.03503685376608

Determinism

hurst_exp = signa.determinism(x)
print(hurst_exp)

Output:

4.315972222222222

Alpha amplitude

hurst_exp = signa.alpha_amplitude(x)
print(hurst_exp)

Output:

0.19518190403498442

Hurst Exponent

hurst_exp = signa.hurst_exponent(x)
print(hurst_exp)

Output:

0.5074992385199263

Extract all feautres

Instead of use a list indicating the features, you can use the parameter all to calculate all the features.

all_features_calculated = signa.extract_features(x, 'all')
all_features_calculated

Output:

{'mean': 0.006880339575229891, 'std': 0.9641841680210175, 'coeff_var': 0.9296511098623816, 'median': 0.07802095042293272, 'mode': 0.47143516373249306, 'max': 2.390960515463033, 'min': -3.5635166606247353, 'first_quartile': -0.4874633299633267, 'third_quartile': 0.6768295649371497, 'inter_quartile_range': 1.1642928949004765, 'kurtosis': 0.9355744131054928, 'skewness': -0.5940081255524681, 'activity_hjorth_param': 0.9296511098623816, 'mobility_hjorth_param': 1.5772426970445614, 'complexity_hjorth_param': 1.1381489369369338, 'total_power_spectral_density': 1.03503685376608, 'centroid_power_spectral_density': 39.785545239633095, 'relative_delta_power': 0.014199533696614332, 'relative_theta_power': 0.024773922507739236, 'relative_alpha_power': 0.06670507211034518, 'relative_beta_power': 0.16222741368578988, 'relative_gamma_power': 0.1944461143387118, 'determinism': 4.315972222222222, 'trapping_time': 2.2205882352941178, 'diagonal_line_entropy': 2.794104878439014, 'average_diagonal_line_length': 26.28700906344411, 'compute_recurrence_rate': 0.063720703125, 'spectral_edge_frequency_25': 27.0, 'spectral_edge_frequency_50': 46.0, 'spectral_edge_frequency_75': 55.0, 'delta_amplitude': 0.332883261842357, 'theta_amplitude': 0.1334128596782051, 'beta_amplitude': 0.44829663314331475, 'alpha_amplitude': 0.19518190403498442, 'gamma_amplitude': 0.36127348831240264, 'hurst_exponent': 0.5074992385199263}

Convert the features into a feature vector

feature_vector = list(all_features_calculated.values())
print('features:', feature_vector)
print('n_features:', len(feature_vector))

Output:

features: [0.006880339575229891, 0.9641841680210175, 0.9296511098623816, 0.07802095042293272, 0.47143516373249306, 2.390960515463033, -3.5635166606247353, -0.4874633299633267, 0.6768295649371497, 1.1642928949004765, 0.9355744131054928, -0.5940081255524681, 0.9296511098623816, 1.5772426970445614, 1.1381489369369338, 1.03503685376608, 39.785545239633095, 0.014199533696614332, 0.024773922507739236, 0.06670507211034518, 0.16222741368578988, 0.1944461143387118, 4.315972222222222, 2.2205882352941178, 2.794104878439014, 26.28700906344411, 0.063720703125, 27.0, 46.0, 55.0, 0.332883261842357, 0.1334128596782051, 0.44829663314331475, 0.19518190403498442, 0.36127348831240264, 0.5074992385199263]

n_features: 36

Development

SignaLytica was created and is maintained by Edgar Lara. Contributions are more than welcome so feel free to contact me, open an issue or submit a pull request!

To see the code or report a bug, please visit the GitHub repository.

Note that this program is provided with NO WARRANTY OF ANY KIND. Always double check the results.

Acknowledgement

This package was inspired by the research work of Hernández Nava, G. (2023). Predicción de eventos epilépticos mediantes técnicas de aprendizaje profundo usando señales de EEG.