[book review] The hundred-page machine learning book
I have mixed feelings about this book. Let’s start by looking at what was promised by the author.
Andriy Burkov claims that the book is sufficient for beginners and gives them “enough details to get to a comfortable level of understanding.” He also promises, “a collection of directions for further self-improvement” for experienced practitioners. All of that in one book that was supposed to be hundred-page long. Is it even possible?
On the one hand, I think that Andriy failed to achieve all of his goals. First of all, the book is 50% longer than promised, but that is fine. After all, the “hundred-page” was just a marketing trick.
What about other promises? He fails to deliver value for beginners. In my opinion, they will get lost, confused and overwhelmed by the subject. There are no examples that may help readers understand the concepts.
Experienced machine learning practitioners won’t find anything new in this book. I think that readers who are familiar with the matter will be disappointed by the focus on the theoretical aspect of machine learning and the lack of useful hints.
On the other hand, Andriy’s book is perfect for people who want to get familiar with the mathematics behind machine learning algorithms. It gives the readers just enough details to understand the idea and skips everything that is not crucial.
That is the only thing I liked about this book. Every sentence in this book is essential. The content is extremely dense. The author is not beating around the bush. There is nothing that can be left out.
There are no cute stories, private opinions, use case examples or anecdotes. Every paragraph contains either a definition of something or an equation. For sure, it is very informative, but is it useful? I highly doubt it.
For me, this book is like a dictionary. It won’t give us everything we need, but it is excellent as quick reference material. The problem is, no dictionary will help us solve a real problem. “The hundred-page machine learning book” may be a convenient starting point when we look for a solution, but so is Wikipedia.
Do I recommend the book? No, I don’t. Don’t get me wrong. The book is awesome. If I had to pass an exam, it would be a perfect book for studying. It just won’t help me solve any of my machine learning problems. That is why I cannot recommend it.
Did you enjoy reading this article?
Would you like to learn more about software craft in data engineering and MLOps?
Subscribe to the newsletter or add this blog to your RSS reader (does anyone still use them?) to get a notification when I publish a new essay!
You may also like
- How to be happy at work - lessons learned from "Career superpowers" book
- 5 best books I read in 2018
- [book review] Deep work by Cal Newport
- "The war of art" and other books I did not finish reading
- "The Box: How the Shipping Container Made the World Smaller and the World Economy Bigger" by Marc Levinson
- Data/MLOps engineer by day
- DevRel/copywriter by night
- Python and data engineering trainer
- Conference speaker
- Contributed a chapter to the book "97 Things Every Data Engineer Should Know"
- Twitter: @mikulskibartosz