License and citation#

Code#

The code of pyml-datasets is released under the MIT license, see the LICENSE file in the repository and in the installed package metadata.

Data#

The datasets were created by other people, so each one keeps the license of its original source. Only two licenses are accepted in this project:

  • CC0 1.0, a public domain dedication. Nothing is required, but citing the creators is good practice.

  • CC BY 4.0. You may use and share the data, also commercially, if you give credit, link the license and say if changes were made. The citation on each dataset page and in get_info(name).citation is the credit, and the changes are listed on the same page.

The license texts are in the LICENSES folder of the repository and of the installed package, together with NOTICE.md, which lists the creators, sources, citations and changes of every dataset.

How a dataset is admitted#

A dataset is bundled only if its source states one of the two licenses above. A dataset with no license, with a non-commercial or share-alike license, with an “educational use” statement or with an unclear origin is not bundled, even if it is very well known. For each dataset the repository keeps a short evidence file (legal/evidence/<name>.md) with the page that states the license and the date it was checked, and the automated check python -m scripts legal fails if any dataset lacks one. The rules are written in the contributing guide of the repository.

Important

I am not a lawyer and this is not legal advice. The licenses were read at the source on the dates in the evidence files. Licenses can change, so check the source yourself if it matters for your use.

Removed datasets#

Version 0.1.0 also contained Diabetes, Linnerud and California Housing. Their sources do not state a license that allows redistribution, so they were removed in 0.2.0, see the Changelog.

Citation#

If you use a dataset in published work, cite its original authors. The references are in the Source and citation section of every page in Datasets.

Takedown and corrections#

If you hold the rights to a dataset here and the license, the attribution or the changes listed are wrong, or you do not want it distributed, write to juraevsherzod9943@gmail.com. It will be corrected or removed.