tsxtract

tsxtract.extraction

Feature extraction functions.

tsxtract.extraction.extract_features(dataset)

Extract features using tsxtract.

Parameters:

dataset (jax.Array) – Dataset to extract features from. Must be an array of shape (samples, channels, length).

Returns:

Dictionary with feature names as key and extracted features as values.

Return type:

dict[str, jax.Array]

tsxtract.extraction.maximum(signal)

Get the maximal value in the signal.

Return type:

Array

tsxtract.extraction.mean(signal)

Calculate the mean value of the signal.

Return type:

Array

tsxtract.extraction.minimum(signal)

Get the minimal value of the signal.

Return type:

Array

tsxtract.utils

Utility functions for tsxtract.

tsxtract.utils.generate_random_time_series_dataset(n_samples=100, n_channels=5, sampling_rate=100, time_series_length_in_seconds=10.0, *, random_seed=None)

Generate a random time series dataset.

Parameters:

• n_samples (int, optional) – Number of samples to generate, default is 100.

• n_channels (int, optional) – Number of channels per sample (1 = univariate, >1 = multivariate), defaults to 1.

• sampling_rate (int, optional) – Sampling rate in Hz, default is 100 (Hz).

• time_series_length_in_seconds (float, optional) – Length of each time series in seconds, defaults to 10.0 seconds.

• random_seed (int | None, optional) – Random Seed for reproducibility, defaults to None, which equals to random_seed = 0.

Returns:

Randomly sample time series dataset with shape (n_samples, n_channels, int(sampling_rate * time_series_length_in_seconds))

Return type:

jax.Array

Notes

The data is sampled using a normal distribution.

Examples

>>> from tsxtract.utils import generate_random_time_series_dataset
>>> array = generate_random_time_series_dataset(100, 3, 10, 10)
>>> array.shape
(100, 3, 100)