When the classic ML textbook skips the algebra

What it does 南瓜书 (Pumpkin Book) is a companion reference to Zhou Zhihua’s widely used Chinese textbook Machine Learning—nicknamed the “Watermelon Book.” Where the original omits derivations to stay accessible, this project fills in the missing algebra and explains difficult formulas from an undergraduate math perspective. It is essentially a set of community-curated study notes for readers who lack the “solid sophomore-level math foundation” the original author assumed.

The interesting bit The authors originally assumed the textbook omitted details to be beginner-friendly, only to learn that Zhou Zhihua expected readers to derive the gaps themselves. The entire project is therefore a humble admission that most of us cannot mentally fill in the missing algebra, and that is okay.

Key highlights

• Covers formula-by-formula derivations for the 2016 edition of Machine Learning
• Explicitly targets readers with standard undergraduate math, offloading advanced topics to appendices
• Maintains a 24-hour response pledge for formula requests and errata via GitHub Issues
• Published a physical edition through People’s Posts and Telecommunications Press, with the open-source repo remaining as the raw draft
• Licensed under CC BY-NC-SA 4.0, with accompanying video lectures and study groups

Caveats

• Content assumes you are actively reading the Watermelon Book; it is not a standalone introduction
• The authors advise beginners against deep-diving into Chapter 1 and 2 formulas early on
• The open-source version is an unedited draft; the printed book contains additional editorial polish

Verdict Grab this if you are working through Zhou Zhihua’s textbook and keep getting stuck on the “trivial” algebraic gaps. Skip it if you are looking for a first machine-learning primer or already derive every omitted step without breaking a sweat.