Time-based k-fold validation splits for heterogeneous data.

Folds plotted on a two-by-two grid. See the examples page for more.

A library for creating time-based cross-validation splits of heterogeneous data, such as raw transaction data with strong non-stationary characteristics.

• Splitting schedules based on a fixed interval, a CRON-expression, or a pre-defined list.
• Data selection based on a timedelta, or the splitting schedule itself.
• Automatically extract and normalize data limits for supported data types.
• Plotting function for visualization of folds. Display fold sizes (hour/row count), or use custom text.
• Integrations for popular libraries such as pandas, polars and scikit-learn.
• Convenient web application for exploring folds with different parameters.

The Time Fold Explorer application (available here) is designed to help evaluate the effects of different parameters. To start it locally using Docker, run

docker run -p 8501:8501 rsundqvist/time-split

in the terminal. You may use create_explorer_link() to build application URLs with preselected splitting parameters.

The package is published through the Python Package Index (PyPI). Source code is available on GitHub: https://github.com/rsundqvist/time-split

pip install -U time-split

This is the preferred method to install time-split, as it will always install the most recent stable release.

If you don't have pip installed, this Python installation guide can guide you through the process.

MIT

Hosted on Read the Docs: https://time-split.readthedocs.io

All contributions, bug reports, bug fixes, documentation improvements, enhancements, and ideas are welcome. To get started, see the Contributing Guide and Code of Conduct.