The Hundred-Page Machine Learning Book

The "Hundred-Page Machine Learning Book" by Andriy Burkov, is in my opinion the best book for those working with machine learning libraries but don't have an understanding of the underlying science behind the libraries. I am a machine learning scientist/ engineer and often get asked what is the difference between what I do and what someone that just applies libraries are. This book explains it in a very down to earth way. Yes in this book there is some math used, nothing too excessive and should be easy for anyone with some mathematical experience to grasp.The best part of the book I have to say is that it gives the introduction that I think so many need in understanding that a simple ML library and a coder are not going to present the answers to the questions being asked. If there is a simple question with a clean dataset, then yes someone with some tech knowledge will be able to grab a ML library and come to a conclusion. For a more in-depth question with a messy dataset or when the basic libraries don't cover the problem someone with more in-depth knowledge of the science is needed. Thank you Andriy Burkov for writing this amazing book.

While this book is presented as a primer for Machine Learning enthusiast and practitioners ranging from beginner to expert, I would argue against diving headfirst into it if you have not done some pre-work first. To be fair, the book really is well put together and does a great job of explaining and connecting various ML concepts. However, to get the most out of this text I recommend the following build-up process before reading.First, read the “Machine Learning for Everyone” post on the vas3k blog to make sure this is a topic you are really are interested in, and to demystify yourself of that the idea that all of this ML stuff is synonymous with AI. Second, take a Python course at either codecademy or a similar online self-paced learning website. Then I recommend taking a Linear Algebra course at a site like Kahn academy, especially if it has been many years since your last math course. Lastly, I recommend taking a Data Science course at either General Assembly, Kaggle, or Data Camp to get the basics and generally terminology down. At this point you will likely be a bit overwhelmed and wondering how to form one coherent big picture of all of this stuff you just learned.It is at this point where the “The Hundred-Page Machine Learning Book” comes in perfectly. It resets all you have learned to this point into one coherent picture. Reading it at this point also will require less look up of the referenced topics, math, and libraries / tools. While the book is only 100 pages and reads fast, it is probably best to chunk it up into no more than a chapter a day and read it over two weeks. It has also has a long shelf life because you will find yourself referencing back to it from time-to-time.

just bought the kindle version and it does not work on my kindle. LOL. Works on my iPhone only and this sucks!

This book is a nice addition to a practitioner library.It definitely needs pre-existing knowledge to fully appreciate it, but it is a nice starting point to know what to google next. It also has a nice online wiki that goes in depth on most topics, better yet, you do not need a code for the wiki. The book is also available online, check:Themlbook.comI recommend not to be cheap and buy the hardcover. It is imperative to support the authors of such initiatives.It covers pretty much any important algo you need to know, still I found it a little bit light on clustering and clustering evaluation techniques.

The Hundred Page Machine Learning Book is a great standalone book as well as a good supplement for other material in your ML library.The author breaks down the concepts in ways that don’t over-simplify the material or present it in a way that assumes you have decades of advanced math under your belt. (He’s found the sweet spot between too easy and too hard; it’s just right in my opinion).The author explains things in an easy-to-understand way, one that facilitates comprehension of the more math-heavy ML academic textbooks. This doesn’t mean his book doesn’t have math. He just does a great job of explaining everything, breaking it down to exactly what you need to know.One thing I appreciate is the author’s “try before you buy it” approach, whereby you can read chapters of his book online (visit the author’s website for more info) before making your purchasing decision. That said, I think if you’re new to ML and planning to learn ML through self-study, then this book is a must. I’d also recommend it to anyone who needs to get grounded before stepping into a ML academic course (or as in my case, as an excellently written, concise supplement to other ML material).I have no hesitation in recommending this book!

I have read 5 chapters and I think this is one of those books that you know you have to buy and keep it when you need a reference. It is well written, and it goes right to the point. I really enjoy the examples and if you want to go deeper it provides some web resources. If you buy a physical copy, I would recommend the hard copy. It is well done and seems durable. Thanks for having put this beautiful book there.

Great for beginners

On pages 4-6 of Kindle version, author describes simplest linear Support Vector Decision algorithm and lists as a one of the constrains "large margin contributes to a better generalization (...) to achieve that, we need to minimize the Euclidean norm of w denoted by ||w||"And even more, provides 2D diagram describing his point; and lists "b/||w||" which is distance to (0,0) coordinates and which is not related.Note that we can infinitely minimize ||w|| (which is "length of a vector" in Euclidean metric space): 32x +32y + 32 = 0 defines the same line as x+y+1=0, and same as 0.1x+0.1y+0.1 =0, and so on. I am wondering if reviewers ( VPs and CEOs of famous companies, some of them I know personally) indeed read this book. Such mistakes are unforgivable, I am wasting time.And Page 8 is empty, and book has 136 pages from Introduction to Index. Page XVIII of Preface mentions "this book is distributed on the 'read first, buy later, principle." almost each page contains noisy barcode pointing to supplemental website which also mention the same "read first, buy later" (and no way you can find this feature available!); no any URL links in Kindle version which I can just point and click; and somehow page navigations are missing (Kindle for Mac). No any "further reading" at the end of chapters. The book could be good introduction for beginners if at least it has really working website with interaction to readers, with electronic version available for free - compare with "Thinking in Java" book (which is free, and which I have paid-for hardcopies of three editions).

I have just started with Data Science since past few months.Never in my life apart from the academic books I have purchased any other book as reading is not my thing. Also I am afraid of those big big books.The 100 page machine learning is the first book that I have purchased in my life.And I am really excited to go through it and start my Data Science journey.Thank you.

A principal vantagem deste livro é que ele de fato pode ser lido de cabo a rabo sem que seja necessário empreender um esforço hercúleo para tanto. Cobre o básico de aprendizado supervisionado e não supervisionado, técnicas de regularização, redes neurais e deep learning (incluindo uma boa explicação sobre redes convolucionais). Eu já havia me iniciado no mundo de Machine Learning através de cursos e outros livros, e devo dizer que este consegue explicar muita coisa de forma simples e direta. Desconheço outro livro que consiga cobrir tão bem tantos tópicos em tão poucas páginas. Leitura recomendada para quem quer se iniciar no mundo do ML ou mesmo ter um livro de referência para consultar sobre o assunto (eu já terminei de ler e continuo usando o livro dessa forma de tempos em tempos).Ps: a versão digital não é para ser visualizada no Kindle, e também não é bem otimizada para tablets, por ser print réplica.

Product quality is so bad.I read the PDF and it had great content but disappointed by paper quality.