Hacker News Reader

Comment by codypo | Hacker News Reader

The Programming Collective Intelligence book on sale here is wonderful! If you want to build software that does something intelligent thanks to the mining of a tremendous set of data (think recommendation engine, price models, content classification, search engine, etc) and you're not sure where to start, start with this book. Not being a CS academic, it's hard to find useful, non-trivial introductions that I can understand on these topics. That's where this book excels. It's one of the few technical books I regularly go back to.

View on HN

Seconded. It's got some minor errors* and omissions, but after reading it for the concepts, I've found other data mining / machine learning books much easier to follow. The explanations are usually pretty good.

* I've been doing the exercises in Lua, and got stuck on one of them - the listed algorithm (IIRC) erroneously ended with "1 - num/denom" rather than "num/denom". It's listed correctly earlier. (I can look up the specifics later, I don't have the book with me.)

Off and on I've been doing the exercises and examples in Clojure, and it's a little difficult because the Python examples seem so imperative to me. It's been a challenge turning them into functional code, there's a couple of places where I go over a list twice when I probably should be able to get all I need with one pass, and of course when I have to resort to using an atom or ref I feel a little dirty.

Thanks! I'm the author of this book, and it always makes me so happy whenever I read a comment like this on HN :)

I hate to "met too" but I'm a big fan of it as well! I've built a few of the python examples into their perl equivalents over the last few months. Someday I should throw 'em up on github.

Thank you, this answers pretty much all the questions I had for the book. Just added to my cart, and now I`m trying to decide on the "MongoDB: The Definitive Guide" book.

I agree that PCI is a great introduction to the field. If you're not sure which tool to use to solve a problem, it's great for pointing you in the right direction. Just a word of warning (@toby, correct me if I'm wrong), but most of the examples are just to give you the idea and to get you started, but aren't expected to work at scale. I'd love to see a followup book on "Optimizing Collective Intelligence".

This will be extremely helpful while reading the book. I hope there will be an updated version in the future, one of the best introductory books I found on this subject.

http://oreilly.com/catalog/errataunconfirmed.csp?isbn=978059...

I'd like to add that Stephen Marsland's 'Machine Learning: An Algorithmic Perspective' is a new book in the same practical/introductory class. The code is mostly numpy.

http://www-ist.massey.ac.nz/smarsland/MLbook.html

Reply on news.ycombinator.com