// the find
amueller/scipy-2016-sklearn
Scikit-learn tutorial at SciPy2016
A Jupyter notebook tutorial from SciPy 2016, covering scikit-learn from basics through pipelines, cross-validation, SVMs, and out-of-core learning. Written by Andreas Mueller (a scikit-learn core contributor) and Sebastian Raschka. This is a teaching artifact from a conference 10 years ago, not a library or tool.
The curriculum sequencing is genuinely good — it builds from data representation through model evaluation before going deep on individual algorithms, which is the right order. The inclusion of solutions for every exercise makes it self-contained for self-study. Mueller's authorship means the API explanations reflect how the library was actually designed to be used, not how beginners assume it works. 23 notebooks covering the full supervised/unsupervised split, plus pipelines and text classification, is more thorough than most intro ML courses.
Abandoned in 2019, targeting Python 2.7/3.4/3.5 — scikit-learn's API has changed enough since 2016 that several patterns here are deprecated or wrong. No coverage of anything post-2016: no ColumnTransformer, no HistGradientBoosting, no SHAP or model explainability. The YouTube videos linked in the README are the actual teaching mechanism and those won't age well as a reference. Anyone picking this up today is better served by the official scikit-learn tutorials, which are actively maintained.