Thanks for initiating what I hope will be a very interesting discussion!
I believe most people, including data scientists, don't make full data-driven decisions about cars. Emotions associated with things like style preferences tend to drive decisions and choice of colors. BTW, I learned that for a given model, year, and mileage, used care resale prices are better for black, white, and silver cars.
You might start by listing the factors that mean the most to you and then weight them. Include things like style along with qualitative and quasi-quantified factors like ride comfort, noise, and reliability found in Consumer Reports, for instance. Rate a few of your top possibilities according to this method and see how they do.
As a second step, perhaps as a sanity check, you might estimate the total cost of ownership for your top choices. I think this is a good measure, since it takes into account things like purchase price, resale value, financing cost, operating cost, insurance, property taxes, and maintenance. Most of these are readily available data, except maintenance. I personally like to buy an extended warranty on my cars, usually for 7 years and 100,000 miles. I know this will cost me more than I would likely pay for maintenance, but I feel it has peace-of-mind value. The cost of such warranties varies significantly by manufacturer and model. I think it's a good indicator of expected maintenance costs.
After combining these two views, it will be interesting to see if data makes a difference in your final decision.
------------------------------
Norm Fujisaki
Broadlands VA
------------------------------