Originally Posted by TboneZX3
You would have to assume that they trend one way or another throughout the graph though (city vs. highway). What if, for one the drivers, he drove the first half of his fillups on the highway, and on the last half of his fillups he drove in the city? ...or some other combination. Would that be good data for your study?
You are correct. I am assuming those drivers are few and far between and won't have a huge effect on the overall gist of the graphs.
For example, that driver in my second graph who increased 13 mpg over time, that's probably someone you're talking about. But clearly he/she is an outlier.