This book has taken me longer to read than just about any other technology book, but I’ll come back to that.

The premise of this book is that machine learning has become an integral part of modern applications, from Amazon’s product recommendations through Google’s pagerank and on to decision making, filtering and clustering and a few others in between. Basically a range of different approaches to extracting meaning from data. Each chapter covers a different technique from a very practical angle: you actually build an implementation of the algorithm in question.

These topics could easily ascend into a theoretical discussion, but the author does a surprisingly good job at keeping things practical and grounded, and therein lies the reason I’ve been reading this book for several months. He keeps it so practical and interesting I’ve felt compelled to actually build every example. Along the way I’ve gone from having a fairly dusty Python knowledge to being very comfortable in the language, an unexpected side benefit.

It’s not easy to make a discussion about an algorithm interesting. Julian Bucknall, both in his columns in The Delphi Magazine and now PC Plus, and also his Algorithms and Data Structures book, is still probably my favourite, however Toby Segaran does a pretty effective job as well.

Don’t buy this is you want a high level tour of machine learning. However if you really want a deeper understanding and you’re prepared to roll your sleeves up and write code in order to get it, potentially learning a new language along the way, I can thoroughly recommend this book. Just don’t expect to read much else for a little while.