How many SkyActiv gasoline and diesel Mazda 3s and CX5s are listed on
fuelly?
GO to fuelly and look at samples that are 100,000 miles or more of fuel logs, not estimates.
That does not necessarily make the ACCURACY better than "USER ESTIMATE". It just has the potential of more data points with greater uncertainty than those pre-screened for unreasonableness prior to inclusion in EPA's
User MPG Estimates.
Keep in mind that the term "Estimate" is the result of all of the variability between drivers, environments, topology, and driving conditions (road, weather, and trafic).
Simply put ... there is NO ... SINGLE ... mpg value ... that fits ALL drivers in all conditions for a given vehicle model. Hence ... "Estimate".
As for the estimate 8% above sticker ... go look at the >2.5 Liter Det3 gasoline machines and see how many "USER ESTIMATES" reflect >8%
below sticker mpg combined. I think you will find the range generally between 8% and 15%
below combined sticker.
If you are really interested ... here is a technial paper to study on the subject:
SAE Technical Papers #2011-01-0618 © 2011: "Predicting Individual Fuel Economy" by Lin, Z., and Greene, D. Oak Ridge National Laboratory
BTW, I have seen documentation indicting that there are far more than 35K vehicles in the "USER ESTIMATES" database accumulating mpg data, some at rates of 12K/year or more, some for many years. A good example is the Honda Civic, with multiple annual configuration going back to 1987 (25 model years).
sky ... I'll let you count the number of samples and data points for the Civic.
Don't get me wrong ... fuelly data is OK ... but, one would have to do a lot of statistical work on the data to approach the quality of EPA's Shared MPG Estimates, IMO.
However ... it is your choice ... I'm just sharing some of the "stuff" I have learned over 7 years of study ... take it or leave it ...
HAPPY MOTORING ....