Machine Learning in Action

by Peter Harrington

The ability to take raw data, access it, filter it, process it, visualize it, understand it, and communicate it to others is possibly the most essential business problem for the coming decades. "Machine learning," the process of automating tasks once considered the domain of highly-trained analysts and mathematicians, is the key to efficiently extracting useful information from this sea of raw data.

Machine Learning in Action is a unique book that blends the foundational theories of machine learning with the practical realities of building tools for everyday data analysis. In it, the author uses the flexible Python programming language to show how to build programs that implement algorithms for data classification, forecasting, recommendations, and higher-level features like summarization and simplification.

Genres: Programming, Computer Science, Artificial Intelligence, Computers, Technology, Technical, Mathematics

384 pages, Paperback

First published December 28, 2011

About the author

There are several authors with this name on Goodreads.

Reviews

[Andre · Rating 4 out of 5]

Want to know where, in Portland, OR, to park so that you can walk to the most strip clubs?

Yes, this is a real example in this book (the data set consists of Magic Gardens, Mary's, Dolphin II, etc). I kid you not. As a result, I'll never forget the k-Means algorithm.

[Rex · Rating 5 out of 5]

At first I just liked this book because it had some nice explanations about the basics of machine learning and I was interested in a general overview. Then I encountered single value decomposition and latent semantic analysis and soon found that the Internet only contained dozens of horrible purely academic explanations about the underlying math that were basically impenetrable. This book, on the other hand, had a very clearly worded walk through on a topic that is otherwise scarily difficult to find.

[Sebastian Gebski]

Unfortunately, I can't provide a discrete rating. Due to following reasons:

* usually when things started to get interesting (for a particular algorithm / method) instead of diving into mathematics behind method, author was jumping into Python; for majority of people it really makes sense -> developers are not mathematicians, but I personally dislike Python, so I'd rather prefer some more details so I could more easily map them to other programming language; but I can't blame anyone but myself - author didn't hide the fact that code samples are in Python

* due to other stuff I was doing in the meantime, I wasn't able to do any practice alongside reading the book - and it's a problem, because it's one of the books you can't just read without putting your hands on code; You have to actually touch the described algorithms / methods to make some sense out of them - good example: Adaptive Boosting; sadly I've failed again - I didn't have time for that :(

Anything else to add?

Well, apart from what I've written above:
1. There's a section (3 chapters) about Unsupervised Learning - it's great, because I haven't seen many practical (non-pure statistics) books on the topic recently
2. Some chapters seem a bit out of topic - the Hadoop stuff, Tkinter.

To summarize - I can't give a 'star' rating. And I have a feeling I'll be coming back to this book in a bit more convenient time.

[Kursad Albayraktaroglu · Rating 5 out of 5]

It's been three months since my wife asked me why I was "always reading the same book with the axe guy on the cover"; but I am finally done with this book. It took a lot of effort to finish it by working out all the examples and exercises; but I think it was well worth it. I personally think Harrington's book strikes a very good balance between mathematical and programming aspects of ML; and would be a great introductory book for anyone with working knowledge of Python and preferably some background in statistics.

Since the book heavily depends on Python code examples and exercises, it may not be the best choice for a non-programmer: the author prefers to explain many complex subjects in code, and you will not understand the material if you skip the coding examples. It looks like Paul Wilmott's new book and Andriy Burkov's "The Hundred-Page Machine Learning Book" are better suited for the more mathematically oriented ML learners. For anyone approaching ML from the software development side, Harrington's book is highly recommended.

[Suhrob · Rating 4 out of 5]

A book caught in the uncanny valley...

Harrington strives to give introduction to basic machine learning topics and algorithms by developing them from scratch in python (using numpy and matplotlib but not scipy/scikit-learn). This way he indeed gives more insight than just a completely black box approach but nowhere near as much understanding as a proper mathematical treatment of the algorithms. On the other hand the implementations are rudimentary and in fact for all practical purposes one would just use any state-of-art library instead the introduced basic algorithms.

So the book is neither theoretical (too shallow for that) nor practical (not showing you any of the libraries actually used in practice).

It is written with earnestness and with more care than what's typical for similar programming handbooks, but I have no idea who I could recommend it due its neither-nor nature....

[Kai Jiang · Rating 4 out of 5]

Read this when second year of my Master's. This book is more like a intro or shallow summary for the huge Machine Learning world, yet it comes with easy understanding Python code and just the right amount of math background so you can get the sense very quickly.

[Avinash K · Rating 3 out of 5]

The book is good for the summary information and to get your feet wet! The writing is lucid and not intimidating. So one can start wrapping one's head around the ideas. On the flip side it lacks the required mathematical background. So the book a starter, a decent starter. A good book to have along side is "The Elements of Statistical Learning" statweb.stanford.edu/~tibs/ElemStatLe...

[Kiril Kirilov · Rating 2 out of 5]

I personally think that the Coursera's course is much better way to inform the unprepared mind about the marvelous world of Machine Learning algorithms.

[Cheogm · Rating 4 out of 5]

I really love this book but I can not give it a 5, because to fully understand the topics I had to study from other media the mathematics it is not like math and code for example sometimes the author develops an algorithm which has more steps from what is develop and what is presented as the equation. By the other hand it is a very good way to start the topic. I am pretty sure that after you read this book you will understand at Lest when people talk about the topics in a very good and deep way.

[Frank · Rating 2 out of 5]

I don't think this book is for me. It explains algorithms, not that much with math but with code.

[Rick Sam · Rating 4 out of 5]

A Top Down Approach to Machine Learning with code



Deus Vult,
Gottfried

[Matija · Rating 2 out of 5]

I wouldn't recommend this book. I read parts of it a few years ago before I knew much about either machine learning or Python, and it didn't help me much - it provides only rudimentary and descriptive information on algorithms, and uses Python in a hacky way (importing *, repurposing libraries, etc.). After rereading it now that I know more on both subjects, I actually see some errors and bad practices in it too. As another reviewer, @Suhrob accurately writes, state-of-the-art algorithm implementations already exist in libraries for Python. Instead, the author provides his own incomplete implementations without enough solid background to justify the whole exercise. The only use I see for it is as a quick reminder on basic descriptive facts about a few machine learning algorithms.

[Phil Sanders · Rating 3 out of 5]

After taking the free Stanford class on machine learning, I was we'll-versed in the theoretic math with very little practical knowledge. I felt this book filled that niche and, for the most part, was very readable. On the downside, I think the annotated listings could have been better and a couple of the examples kind of left me hanging and seemed incomplete.

[Michal Paszkiewicz · Rating 4 out of 5]

A great introduction to machine learning algorithms and implementations. Personally I would have preferred a bit more maths though.

[Frank Palardy · Rating 4 out of 5]

This series they like to come up with complicated examples but machine learning is complicated enough already.

[Rui · Rating 4 out of 5]

Good book overall.

I hope the author has put more effort to the implementation code.
Expect to spend some time debugging the code.

[Bahrun Nur · Rating 4 out of 5]

This is a good book for beginner that wanna know and try Machine Learning algorithm.