Member-only story
The 6-Minute Guide to Scikit-learn’s Version 1.0 Changes 😎
A quick tour of the greatest hits 🎸
Scikit-learn has been the primary Python machine learning library for years. The move to version 1.0 doesn’t bring many major breaking changes, but it does include bug fixes, few new features, some speedups, and a whole bunch of API cleanup. 🧹
A more consistent API is wonderful for both new and existing users. In this article I’ll dig into the changes so you know how to take advantage of version 1.0. 🎉
New features
Now scikit-learn let’s you create B-splines with the preprocessing.SplineTransformer. I think of splines like more fine-grained polynomial transformations. As seen in the plot below, splines make it easier to avoid the ridiculous extrapolations you often see with high-degree polynomials.
James et al. are all about splines in their recently updated machine learning touchstone An Introduction to Statistical Learning, 2nd Edition.
