Time Series Feature Extraction Library

Intuitive time series feature extraction

TSFEL is an open-source Python library for time series analysis. It centralizes a large and powerful feature set of several feature extraction methods from statistical, temporal, spectral, and fractal domains.

The documentation is available here.

You can install TSFEL via pip using the following:

pip install tsfel

A release on conda-forge is coming soon.

Getting started

Below is a quick example of how to use TSFEL for time series feature extraction:

import tsfel

# Loads a 10 s single lead ECG
data = tsfel.datasets.load_biopluxecg()

# Set up the default configuration using using the statistical, temporal and spectral feature sets. 
cfg = tsfel.get_features_by_domain()

# Extract features
X = tsfel.time_series_features_extractor(cfg, data)

For a more detailed walk-through — including input/output data formats, extraction routine configuration, and how to implement your custom features — refer to the documentation here.

Highlights

• Intuitive, fast deployment, and reproducible: Easily configure your feature extraction pipeline and store the configuration file to ensure reproducibility.
• Computational complexity evaluation: Estimate the computational time required for feature extraction in advance.
• Comprehensive documentation: Each feature extraction method is accompanied by a detailed explanation.
• Unit tested: We provide an extensive suite of unit tests for each feature to ensure accurate and reliable feature calculation.
• Easily extended: Adding new features is straightforward, and we encourage contributions of custom features to the community.

Available features

TSFEL automatically extracts more than 65 distinct features across statistical, temporal, spectral, and fractal domains.

Statistical domain

Temporal domain

* Disabled by default due to its longer execution time compared to other features.

Spectral domain

Fractal domain

Fractal domain features are typically applied to relatively longer signals to capture meaningful patterns, and it's usually unnecessary to previously divide the signal into shorter windows. Therefore, this domain is disabled in the default feature configuration files.

Support & General discussion

For bug reports, please use the GitHub issue tracker. To make feature requests, share ideas, engage in general discussions, or receive announcements, you're welcome to join our Slack community.

Citing

If you use TSFEL in your work, please cite the following publication:

Barandas, Marília and Folgado, Duarte, et al. "TSFEL: Time Series Feature Extraction Library." SoftwareX 11 ( 2020). https://doi.org/10.1016/j.softx.2020.100456

Acknowledgements

We gratefully acknowledge the financial support received from the Center for Responsible AI and the Total Integrated and Predictive Manufacturing System Platform for Industry 4.0 projects.