Wharton's Pete Fader would agree with you somewhat that big data-style methods are overblown: http://www.datanami.com/2012/05/03/wharton_professor_pokes_hole_in

Wharton's Pete Fader would agree with you somewhat that big data-style methods are overblown: http://www.datanami.com/2012/05/03/wharton_professor_pokes_hole_in_big_data_balloon/ http://www.datanami.com/2012/05/03/wharton_professor_pokes_h... he also has some very cool uses of dirichlet processes (modelling entire competitive industries)