In the United States fuel efficiency for cars is typically measured in miles driven per gallon of fuel consumed (miles per gallon, MPG). This is not true in some other parts of the world where fuel efficiency is measured as the amount of fuel consumed while travelled a fixed distance (e.g. Gallons of gasoline consumed per 100 miles travelled).
If we fit a regression model of MPG vs. Log(Weight) and Drive Ratio, the plot of Externally Studentized Residuals (ESR) vs. Predicted Values contains a bulge in the middle that indicates a violation of the assumption of linearity of the model. This violation can be corrected by transforming the dependent variable, MPG, to 100/MPG, or "Gallons consumed per 100 miles". Another plot of ESR vs. Predicted Values for the new model shows that the problem has been corrected.