The 2017 textbook code that outlived Theano

What it does

This repository is the official code archive for Python Machine Learning, 2nd Ed., a 622-page Packt textbook published in 2017. It houses sixteen chapter-by-chapter Jupyter notebooks and corresponding .py scripts that walk through machine learning with scikit-learn and deep learning with TensorFlow. The author is explicit that these are companion files—the notebooks contain images and headings, but the formulae and explanatory text live in the book itself.

The interesting bit

The repository captures a specific pedagogical pivot: the first edition’s Theano-based deep-learning chapter was torn out and rewritten around TensorFlow after Google open-sourced it in late 2015. It is also unusually candid about its own shelf life, bannering a link to the third edition that superseded it in December 2019.

Key highlights

• Sixteen chapters from data preprocessing and ensemble methods to CNNs and RNNs.
• Deep-learning content rebuilt around TensorFlow, with new chapters on mechanics, image classification, and sequential modeling.
• Includes practical tangents like embedding a model in a web app and handling imbalanced datasets.
• Notebooks are designed for stepwise execution; .py exports exist but are secondary.
• German and Japanese translations are catalogued in the repository.

Caveats

• The author explicitly warns that the notebooks are not useful without the book’s descriptive text and formulae.
• A third edition has been available since December 2019, making this repository a legacy archive.

Verdict

Useful if you are holding the second-edition paperback or need a curated, 2017-era walkthrough of scikit-learn and early TensorFlow. If you want the current syllabus or self-contained tutorials, the author already points you to the third edition.