On Thu, Jan 19, 2012 at 6:04 PM, Daniel Korzekwa <[email protected]> wrote: > It looks similar to the bayesian model I presented below for predicting my > favorite drink. I think that predicting my favorite drink could be > interpreted as a model/content based recommendation similar to : from > Mahout in action, chapter 4.7.2:
Yes though this is talking about a recommender problem, where you're estimating likelihood of liking some soda that the user has *never* bought. I didn't think that was what you were trying to do. If you want to reason that users who drink Pepsi and Barq's but have never bought RC Cola would probably buy it, then yes you can deploy recommenders. > Pepsi is like a Led Zeppelin, a can of Pepsi you can buy is like a Led > Zeppelin record. Taste of drink is like a music genre, e.g. Pepsi and Coke > taste similar. Same comment, this is trying to deliver stuff you've never bought, not find out the best among the things you have. For example, this technique won't decide whether you like Pepsi, Coke or RC best from your purchases. It would perhaps help you figure out that those who like Coke Zero and Coke, but have never bought Diet Coke, might buy Diet Coke. Whatever works for you works, just have a think about whether the question you are asking is really of this form or not.
