Thoughtful Machine Learning with Python: A Test-Driven Approach

Excellent overview illustrated by nice examples. Great introduction.

I wanted to love this book: incorporating testing into machine learning, and using Python to do it! That's perfect!!Unfortunately, the author glosses over some fairly complex machine learning concepts - which is fine if you already know them, but if you did, why are you getting the book? - and he also has many typos, errors, and at times a somewhat scattered writing style. The errors are the really killer issue, though, as some of those errors are with the code. Some of the code does not compile, and there are variables that he references without ever defining them, so I don't think it was due to mistakes on my end.Yet, the topic is perfect and the author does seem to know his stuff. This seems like the sort of book that could be great when the second version comes out, if the author can (1) fix the errors and clean up the writing style, and (2) and more depth on the topics, concepts, and why he is setting up the testing in the manner that he is.

Forget the code, the mere fact that each algorithm is explained I think is what is really valuable, hopefully a second edition is in the works

This provides an excellent overview of the concepts with enough code examples to see what the expected outcomes are.

This is a nice attempt, but it fails in two regards. First, the combination of software engineering (testing) and machine learning does not give any additional value. Second, the book is too short to give a good explanation of all the machine learning algorithms that are presented. This is not helped by poor editing. For example, on page 91 above, it appears that either the e_i or the s_i are morphed into x_i. On the same page, the reason is "D-separability" which the author then does not discuss much (in fact, not at all) leaving the reader bewildered. Unfortunately, there is more than one example. (Including one were the same figure 7.6 appears twice with different captions). The value in the book are the many python programs, but they by themselves are not useful to learn. Besides, since there is a module in python for almost anything, including neural networks, the text of programs using the modules is not enlightening.It seems to me that the author fell into the trap of wanting to write a short book that covers all important machine learning algorithms and does not use Mathematics. These goals are not compatible. In general, O'Reilly books in my experience are much better written and edited. Let's hope that a new edition is larger and more comprehensible.

I'm torn on this book a bit. I've got limited experience in machine learning, this is to say not none, but also not an expert. I didn't connect well here.The nice thing about this book is it attempts to give an overview of all major types of machine learning algorithms. This is great for people climbing the machine learning curve. The author also places importance on advocating for good coding practices & principles early on, which is also good advocacy for people who may be at the beginning of their programming careers. However, it also assumes a significant mathematical background of the reader, which may not be in the beginner's toolbox. It's not in mine.One of the reasons this book is thinner than say a Pakt tome is the author skips explaining redundant intro stuff-- there's no walkthrough on setting up your python environment. This is appreciated on my part. But this is also where I was displeased with the book. The title was likely promoted by the O'Reilly marketing department because Python is such a popular language. On the cover, the word Python is the largest word. Could anyone say they read this book and learned anything new about Python? Doubtful. And that is not the author's agenda here. It happens to be the language chosen for the examples, but it could have been R or several other choices. The title should be interpreted as, "If you know Python, then the code examples will be familiar to you. But don't open this up expecting a Python coding tutorial."I kind of think the appropriate audience for this book is a very tight niche- those who have significant math training, but for some reason haven't covered the algorithms available in machine learning yet.

Hand full examples of machine learning approach

Code shows error under Python 3, which annoys me. All codes of chapter 4 does not work, and you need to revise it.

Good book with explanations