// the find
amueller/COMS4995-s19
COMS W4995 Applied Machine Learning - Spring 19
Course materials from Andreas Mueller's Applied Machine Learning class at Columbia, Spring 2019. It's a well-structured graduate ML course taught by a scikit-learn core developer, covering the full supervised learning stack with heavy sklearn focus. Useful as a self-study curriculum or reference for ML practitioners who learn better from slides and notebooks than textbooks.
The instructor is a scikit-learn maintainer, so the sklearn-specific content (pipelines, ColumnTransformer, cross-validation patterns) is unusually accurate and reflects how the library is actually meant to be used. Slides are paired with executable notebooks so you can run every example. Coverage is practical rather than theoretical — preprocessing leakage, calibration, and feature selection get full lectures, which most ML courses skip. CC-0 license means you can reuse anything without asking.
Frozen in 2019 — sklearn has changed significantly since then (HistGradientBoosting is now preferred over GradientBoosting, Pipeline API has evolved), so some patterns are stale. No video or audio; the HTML slides lose their presenter context without a recording. Deep learning gets minimal treatment, which was already a gap in 2019 and is a bigger one now. The repo is a static archive with no issues or PRs, so there's no community to ask when something doesn't work.