I think the database needs to be restructured in the following way.

The time per ride data is practically useless. Instead, replace it with an entry for the %highway/street driven for that particular tank of gas entry. This way you and others know more precisely where for each tank entry the average mpg for that vehicle was weighted the most in its spectrum of mpg entries. That is assuming it's lowest mpg entry is roughly equivalent (but not equal) to city driving and the highest mpg is equivalent to highway driving. Of course, the temperature and driving time play a role, but this type of entry is much more informative and quantitative.

Most people would know more precisely what percentage of their driving was city versus highway in each particular tank of gas entry rather than what was their average driving time per ride.