Ask HN: How do technologies like Farecast (now bing) and Flightcaster work?

3 points by ablerman ↗ HN
They take historical data and predict future trends. Does anyone know of a basic introduction to the mathematical models involved? Something simple enough to grasp easily, but with enough depth that the information can be used for small naive experiments?

3 comments

[ 18.9 ms ] story [ 221 ms ] thread
There's really good books on Data Mining at Borders, and a recent bunch of "collective intelligence" books; the Manning books are excellent, but you have know java; you probably also want to install Weka and R, look at the Python suite (numpy, scipy, matplotlib), tools like that; also look up the ~107 (!) algorithms that Bellkor used for Netflix comp.

http://www.research.att.com/~volinsky/netflix/

http://www.amazon.com/Programming-Collective-Intelligence-Bu...

http://www.amazon.com/Collective-Intelligence-Action-Satnam-...

http://www.amazon.com/Algorithms-Intelligent-Web-Haralambos-...

Strong upvote on that book list. If you want to learn from scratch, start with the "Programming Collective Intelligence" book, then move to "Collective intelligence in action", they're both quite good.

If you've covered and understand the material in both, you're probably ready to consider moving to some of the more academic texts.