In progress, still need to find a nice cross section of realistic datasets as I am throwing away the fake data.