Group proposal feedback

[andrewpbray]

Of these three, the first seems like the most promising in large part because you'll likely be able to get access to fairly rich data. The second could conceivable work, but it seems like the data will be less rich. The third I don't think will be particularly interesting because there will be so few predictor variables that you will likely end up with very simple final model.