I was just studying AP Statistics for my exam this week for college credit when this mileage graph idea kind of came to me It sound complicated, but perhaps it could be useful.

You'll see the plotted distance vs. mileage for every car of that model. At the point at which they cross, the mileage can be predicted most accurately. If your car is above the median, you're better than most of your model. If it's above the best fit line, you're better than expected considering your total distance. (Mileage techniques usually improve with driving)

Obviously, the output wouldn't look EXACTLY like this, but what do you think of the concept? It makes for an interesting visual representation.

This can also be done with variables other than distance.

 

cool concept

 

I just realized that point of intersection shouldn't be the point that's most accurate. I *think* if there's an even distribution, it would be the point of average distance of all cars?