Programming Collective Intelligence: Building Smart Web 2.0 Applications
ISBN:9780596529321 read summary

A fun, fast read. Good depth, but still concise. The code is well written, broadly applicable, and easy to modify. This book is the opposite of A New Kind of Science.

This book is spectacular, I love the way that the Author approaches a "new middle" ground of writing books. That is a book that is somewhere in between pure theory, and pure practice. That observation and follow through is simply genius. Python is an excellent choice for this as it can be easy to read. I had to study Python a little before I could totally digest the code. The book is around 300 pages but it is very dense, if someone else wrote it, it would be 600 pages.
Most O'reilly books are boring, useless documentation that you could find on the internet. This book is full of useful examples, showing you how to use "real" data, even how to get the "real" data. For that reason if you are not fond of O'reilly books, don't worry this one is different.

The downside: this book has over 1000 proposed errors and not 1 accepted errors on the O'reilly web site. Some of the code simply does not work as it's written in the book. You can download the code examples but even those do not work 100%. Check the O'reilly site to get the latest code updates. Also the book was published in 2007 and the internet has changed since then so the API's are a little out of date. By not updating this book they are doing the community and they author a huge disservice.

I am not completely finished reading it but I already think it's a great introductory book which is strongly committed with transmitting intuition and comprehension of its material (machine learning) usually hard for regular people. It focuses mainly on implementation and application but some general coverage of the underlying theory is done to motivate inexpert readers. Examples are taken from the Web domain so this text can be very useful for people interested in combining BI and AI, among others. Personally, I approve using Python, which is not (yet) my favorite language, I consider it actually as a plus because it complements very well author's intention of simplicity which is all over behind the book design. This book seems to me like an excellent old school teacher among those ones who really take the right timing and words for carefully explaining you something probably difficult in an easy way so that you really will want to learn more about it.

This book provides good coverage of areas essential to modern web sites. Some complex topics are covered in a manner that anyone can understand. I used this book to suppliment my college material and it helped me understand Gentic Algorithms, Path Finding and other algorithms by giving practicle examples of their use. A must have for any software professional.

It's great to have an accessible algorithms book that doesn't just cover sorting algorithms. This book shows how to implement basic algorithms for clustering, optimization, decision trees etc. For the most part the algorithms are developed from scratch, unlike e.g. in Manning's "Collective Intelligence in Action", which is more of a tutorial on using existing data mining libraries. The drawback is that while the book does a good job of teaching the basics, the algorithms and implementations may be too simplistic for use in a real project.

Python is a good language for code examples, but the code in this book is often a bit too terse (e.g. single letter variables or lists of unnamed parameters), and there are quite a few mistakes in the code.

The "collective intelligence" title and "web 2.0" connection are a bit tenuous. There are some examples where a Web API is used as a source of data, and no doubt some of the algorithms in this book can be useful in such applications. But this is clearly a book about general-purpose algorithms, and not a book about building Web 2.0 sites.