This book is fantastic: It blends theory and practical-implementation seamlessly. There are pure theory ML books, and very practical ML books (such as those from Packt publications), but very difficult to find a book which mixes the two rigorously.
The theoretical part is not abstruse lemma-proof-type, but rather a solid three Chapters on Optimization (with more detailed math left to the Appendix)--more like an Engineering math book; you can derive the equations yourself. The chapters are replete with PYTHON code to help you open up your IDE and crank through. The diagrams/graphics are superlative, very intuitive, and brings home the underlying notions.
What is an absolute gem are the chapters on Feature Learning, Selections and Engineering: Chapters 9, 10 and 11; for that alone, one should purchase this book. Self-consistent, neat and covers nonlinear as well.
The exercises follow the book closely, and in many of them, one has to redo or scenario analyse the material in the body of the chapter, reinforcing your knowledge.