Currently, many people want to learn about Machine Learning. Because they see fancy and intelligent things in the media from big tech companies. To learn about this appealing subject (Machine Learning), there are numerous textbooks and tutorials out there. However, machine learning textbooks are often more than 500 pages. Also, these books are written for the technical audience. That is those readers who have a degree in Computer Science, Mathematics or Engineering. Even CS graduates often find some topics of Machine Learning hard to grasp.
Motivated by the above discussion, Andriy Burkov decided to write a machine learning book that is small and accessible to many people. A while ago, he announced on Linkedin that he was going to write “The Hundred-Page Machine Learning Book”. At first, I thought that this is an impossible task and how he can write a 100-page ML book that covers all the necessary concepts and algorithms. However, he made it. The 100-page ML book is now available for reading. In contrast to what I thought, the book meets almost all of my expectations from a Machine Learning book.
In this post, I review the book “The Hundred-Page Machine Learning Book” by Andriy Burkov. First of all, the author starts the book by telling the truth about the Machine Learning to the readers, which I liked. He states that “Machines don’t learn!”. This is true! Unlike what you see in sci-fi movies, Machine Learning is about mathematical formula and models that maps input data to some output.
Anyway, let’s start the review by the pros and cons of reading this ML book:
- It’s a great book for absolute beginners in machine learning.
- The book covers almost all of the machine learning algorithms that a beginner has to know. Both supervised and unsupervised learning are covered in the book.
- The explanation of concepts and algorithms are clear.
- For most of the algorithms, high-quality Illustrations are provided so that readers can comprehend better.
- Aside from the theory, some useful practical tips such as how to deal with imbalance dataset and missing attributes are given.
- If there were some exercises and computer projects at end of each chapter, readers could practice what they’ve learned.
In summary, I recommend this book to people who are new to the field of machine learning and want to learn moderately the math and intuition behind the ML algorithms. Also, intermediate practitioners in machine learning may learn something new from this book. Finally, if want to read this book, consider buying it to support the author and its work. Because writing a technical book is a non-trivial task and takes a tremendous amount of work and time.
Comments for the author:
You can skip this section of the review. It is written for the author (if he reads this post!). Here are my comments:
- A GitHub repository can be created to host the code samples of the book so that readers can download the whole code samples in a zip file.
- It would be helpful for beginners if there were a section at the end of the book which outlines further actions and what to do after finishing reading this book.
- Reinforcement learning is briefly mentioned in the book and its applications are only stated. In my opinion, at least one algorithm like Q-learning should be demystified.
- In the introduction, SVM is explained to show how supervised learning works. However, based on my teaching experience, SVM is a kind of algorithm that most beginners can’t understand it in the first place (possibly because of its strong mathematical foundation). Perhaps, a simpler algorithm like the kNN might be better for explaining how a supervised method works.
All in all, I’d like to say “congratulations to you”. You did a great job. The 100-page machine learning book helps beginners start out by learning the theory and practice in machine learning. You and I know how frustrating it is for a beginner to study a thick and math-heavy machine learning textbook.