Week 2, 2024: PyLadies talk about custom scikit-learn estimators

This week, I used all my time outside work to brainstorm the details of my thesis topic and work on my Pyladies Berlin talk called “Writing a custom scikit-learn estimator”. The event was yesterday evening, and it was such a great night! Before me, my colleague Stefanie Senger talked about the metadata routing API in scikit-learn, which is a great feature and worth checking out.

On this link you can find the video from the night’s stream, my talk starts the exactly 1:00h mark :). The slides from my presentation you can find here.

Links and resources from the presentation:

• What we do as OSS team at :probabl.
• Documentation about developing a custom estimator
• Link to the validation module documentation, and the source code of the module.
• A video covering check_array() and check_X_y() from the validation module.
• Source code of the estimator_checks.
• Documentation and a list of available estimator tags, and documentation how to use them.
• sklearn-compat, a library to help you support different versions of scikit-learn. Especially relevant after the scikit-learn 1.6 API changes.
• Fairlearn’s website, and GitHub repo.
• scikit-learn certification by :probabl.

That is all for now, see you next week!