Personal preference analysis and recommendation / prediction

UserBird

Hi,



I'm a beginner data analyzer and I'm working on a database which contains user information and the cell phones they've got / liked.

I would like to know if Dataiku can do personal specific recommendations or predictions to each user (like what Amazon does), and if it could, how should I do with it?

Should I use a presiction model (I tried with Random Tree but didn't work so well because there are too many models of cell phones and other related variables, and it generates only one result instead of a list of recommendations ) or should I try other ways ?

I would be appreciated if someone could introduce me how the recommendations are calculated, I am really lost.





thanks!