Hands-On Machine Learning with Scikit-Learn and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems

Hands-On Machine Learning strikes a perfect blend between application and theory. Beginners to machine learning will find it clear to follow and will be able to build complete systems within a few chapters while those with an intermediate level of experience will find a comprehensive, up-to-date guide to this exciting field. Pros: + Practical: The book focuses on examples and implementations of the algorithms rather than the mathematics allowing readers to quickly build their own machine learning models + Readable: Geron does not get too caught up in the details, and he provides warnings when the next section is heavy on theory + Online Jupyter Notebooks: The Jupyter Notebooks that accompany this book (and can even be viewed for free with no purchase from the author's GitHub) are worth the entire purchase price. They feature examples of all the code in the book, plus additional explanatory material. The end-of-chapter solutions to the coding exercises are gradually being added to the notebooks. + Up-to-date: The leading edge of machine learning (and in particular deep learning) is constantly shifting, and Geron does his best to keep the notebooks updated. Multiple times I have read an ML paper and then found the technique implemented in the notebooks within weeks of the publication of the article. Some of the techniques in the book may not be at the absolute forefront of the field, but they are still good enough for learning the fundamentals. + Engaging: The book is a joy to read, and the author is quick to respond to issues pointed out by readers in the book or in the Jupyter Notebooks. Clearly, the author enjoys machine learning and teaching it to others. Cons: - Experts may find this book lacks enough depth because it is more focused on getting up and running rather than optimization. It also is specifically aimed towards Python (and Tensorflow for deep learning) so those looking for implementations in other frameworks will have to search elsewhere. - Due to the rapidly-evolving nature of the field, a print book on machine learning will always need to be periodically re-issued to stay on top of all the developments. Nonetheless, the fundamentals covered in this book will remain relevant and the Jupyter Notebooks are constantly updated with new techniques. Final Line: If you have some basic experience with Python (loops, conditionals, dictionaries, and especially Numpy) and zero to a medium level of experience with machine learning, this book is an optimal choice. I would recommend it both for those wishing to self-study and quickly develop working models, and for students in machine learning who want to learn the implementations of more theoretical coursework. I have enjoyed spending time working through the chapters and the exercises and have found this book extremely useful.

This has to be at the top of my list of most highly recommended books! The amount of material it covers is awesome, and I can find almost no fault with it. The writing is extremely clear, easy to read, written in impeccable English. Very well edited. I don't think I came across any spelling or grammar errors, or any real errors at all. Truly solid writing. The breadth of information covered if quite wide. The choice to start with Scikit-Learn was interesting, but makes sense on some level while he's introducing the more basic machine learning concepts. Simple machine learning techniques like logistic regression, data conditioning, dealing with training, validation, test set. Even if you've read about these concepts a million times, you might still glean useful information from these pages. The Tensorflow section is also super well done. Straightforward setup instructions, pretty intelligible explanation of the basic concepts (variables, placeholders, layers, etc.) to get you started. The example code is quite good, and the notebooks are quite complete and seem to work well, with maybe a few tweaks and additional setup for some. I also found that the notebooks show more examples than what's in the book, which can be nice. I only went really hands on with the reinforcement learning notebook, and found that it was well done and a good base to start my own work from. Even just having a section on reinforcement learning is very rare in a book of this style, and Geron's samples and explanations are really solid. He obviously has a strong grasp of many varied fields within deep learning, and that includes reinforcement learning. The only thing I wish it had was an A3C sample, to make my life that much easier. But you can't have everything. I really liked his tips on which types of layers, activations, regularization, etc. are most effective, and gives good starting points for decent convergence. His explanation of multi-GPU Tensorflow was also quite good. The Tensorboard section was also very useful. In short, if you want ONE book to get you into machine learning, and Tensforlow is on your radar, you can't go wrong with this one. Highly recommended!

It's a good book and I'm enjoying it a lot but there are a few typos or missing function definitions in the code so you need to use the github repository as a supplement when going through this book (unfortunate). The book also includes exercises for each chapter but many of the solutions are just some text on github saying 'coming soon'. A problem with the print book is that the publisher prints in black and white making some of the figures in the print useless, but this is the case with all O'Reilly texts and I wouldn't hold this against the author. I wish I had a pdf to accompany the text to avoid this. Aside from these problems I've found the text to be very insightful and relatively entertaining.

This book has been the greatest resource I've found to learn machine learning. I have intermediate experience with Python and had only a basic understanding of Jupyter notebooks and scientific Python before opening this book, and it's been an awesome experience. This book is: - clearly written - teaches you machine learning through projects rather than jargon - contains a lot of personality (As quoted in a note, "You will often see people set the random seed to 42. This number has no special property, other than to be The Answer to the Ultimate Question of Life, the Universe, and Everything.") - contains exercises at the end of each chapter to recap what you learned This is by no means an exhaustive resource to scientific Python or machine learning, but I've obtained a lot of insight into the methodologies of a data scientist, as well as seeing Pandas methods put into action. Aurélien will provide you a basic understanding of some of the things he does, but it is up to you to use the docs if you want to see the power of some of the methods. With all the things you can do in matplotlib, pandas, numpy, and sci-kit learn, I appreciate seeing them in action, then doing research on them myself rather than sifting through docs to see what methods, functions, or attributes I can use. I have yet to reach the section on TensorFlow, but I'll give an update when I get there. Also, don't be intimidated by the length of this book, it's NOT a textbook.

A very well-written book that takes you beyond the "heavy curiosity" phase of your machine learning education. You need this book if you want to *understand*, from a critical perspective, how to accomplish things like - selecting features, culling data, creating provably-suitable ML models, model/data validation, and what it takes to actually get an ML platform to do something truly useful and meaningful to you! In return for being so useful, the author requests something from you - get your hands on the keyboard, and actually work with Python Scikit-tools as well as the Jupyter workbooks that accompany the book (on github). You should have knowledge of Python, as many of the ML concepts are reinforced through concrete implementations in Python code. And work you must, as - after all - the book's title includes the words "Hands-On"! I respect where the author is coming from, as he is trying to reduce his obvious experience in ML down to a "Hands-On" working environment. I believe he has accomplished this goal very admirably. His easy style of writing encourages you to try things for yourself, and removes the worry that you may "break something" along the way. What truly impresses me beyond even all of this is that English is not his first language! So - again, this work is very impressive on many fronts. SIDE NOTE: This book will likely work for readers with both "step-by-step" and "random-access" learning approaches, as no topic appears to rely so heavily on the previous one(s) that it can't be understood on its own merits. Again - a very impressive feat, especially in light of the heady material being covered. Major Cudos by Mr. Géron!

Excellent book. The information provided is a fantastic introduction to machine learning and deep learning with Tensorflow. I got into the machine learning world via Andrew Ng's Coursera course as well as fast.ai courses on deep learning. I felt like there was always some knowledge assumed about ordinary machine learning that I was lacking. E.g., what's an appropriate cost function? what's the real difference between classification and regression? This book provides that foundation. I finally feel like I have a firm grasp of machine learning. I did not spend too much on the Tensorflow part of the book because I have been working mainly with PyTorch. But whenever it becomes necessary for me to learn Tensorflow, the second part of this book will be the first place I turn. Despite what other reviews say, the print quality of the book is excellent like a typical O'Reilly book. But you have to be careful not to buy it from resellers who will sell you copies of an early release that they printed off the internet and glued together. I made this mistake and had to ask Amazon for a refund to buy the genuine article. The resellers will price their book just a couple of cents below the real thing so when you Add to Cart, it will add their counterfeit book and you won't realize it until you receive it in the mail. Take care to "See all buying options" and only buy if it is Sold by Amazon.com. "Fulfilled" by Amazon is not good enough. The seller itself should be Amazon.com. Otherwise there's a good chance you will get a very poor quality copy that many of the reviews are complaining about.

I am very much a machine learning novice although with a technical background much of the math with its focus on linear algebra and probability and statistics is familiar. I have purchased several other ML books in the past year and have found some better than others. This new book I really like. I downloaded the sample first chapter which is length and a great read and I knew the authors writing style and combination of theory (without a lot of equations) and practice (the hands-on) via Python was just right to encourage me to purchase and read the rest of this book. I am already well into Chapter 3 and I am still very much enjoying this book. Besides the focus on Scikit-Learn, a novel and timely focus in the later part of the book on TensorFlow makes this book a must have if you want to be current. There are a few other titles out there on Deep Learning and TensorFlow but I have not found one that includes an introduce to ML, as well as DL, and these two key Python-based software tools. I definitely recommmend this book as a must have!

I wish I had this book when I was starting out in machine learning and data science. The first few chapters do a great job in giving an overview of basic concepts, friendly overviews of how to munge/preprocess/explore data, and train models. They're not meaty enough to be your only resource for this, but I find it helpful when needing a refresher on a topic and have gotten feedback from colleagues that it gives them a nice introduction to overall topics. The walkthroughs of Pandas, scikit-learn, and Tensorflow are done really well. I have been sitting on getting started with Deep Learning for a while -- none of the data I've examined have been big or thorny enough to warrant it -- but I was building DNNs and benchmarking against traditional models in no time. The supplementary repo is also very useful, but I recommend users develop their own set of functions (e.g. for processing and exploring data quickly, benchmarking models, etc.) on top of it. Along with this book, I recommend Data Science from Scratch for sprucing up on Python and algorithms, Elements of Statistical Learning for a much deeper/textbook-level understanding of concepts, and Data Science from the Command Line for those who prefer CLIs and Bash where useful.

This is a great book if you want to try Deep Learning by yourself. But one warning, don't buy Kindle edition. The mathematical equations are a mess unless you have a big tablets.

By far the most complete and accurate hands-on book on machine learning and deep learning. Author has done a remarkable job in giving details in just the right amount. No over-doing or under-doing in this one. Code given in the jupyter notebooks works like a charm and covers almost everything. Highly Recommended!!

Must read

Dieses Buch ist bei weitem das beste Buch, um Machine Learning mit TensorFlow V1.x zu erlernen. Mittlerweile wird hauptsächlich PyTorch statt TensorFlow verwendet. Und das entsprechende Buch – "Hands-On Machine Learning with Scikit-Learn and PyTorch" [1] – wird noch dieses Jahr erhältlich sein. [1] https://www.amazon.de/Hands-Machine-Learning-Scikit-Learn-PyTorch/dp/B0F2SG98Q9

Having just finished chapter 2, I'm finding this book goes into a lot of detail. Online courses I've taken go through a couple of steps to prepare the data, but this book really takes you through in much more depth. It shows you how to write custom transformers your data and how to make a pipeline and shows you how to evaluate your model as well. I'm really enjoying the book and I think anyone else who knows how to program already and wants to pick up machine learning should definitely pick up this book.