I'm very new to this--I took a class last quarter and got hooked--now I'm in this competition and I don't know where to start (kind of). My problem is that I got this data that looks like a bunch of relational database tables and I don't know how to shape them into something I can try to feed into some kind of model. For example, I have users, users that are friends with other users, events, and users who are attending events. I know how to merge users and events together, convert it to floating point or integer values, and throw it into the model. Its the friends part that's has me confused. How can I bring the friends aspect into the equation?
If your framework will let you represent your data as a directed graph, that might be handy. You could use connections between users to represent a friend relationship, and use a structured machine learning algorithm for your classification. If not, you could create a binary feature vector that's the same length of all the unique visitors with which someone could be friends with, and have a 1 represent friendship, while 0 represents not friendship. Does that help?