[suggestion] filter obviously incorrect data to improve reported vehicle ave mpg

[cajtar]

Note: this idea has been discussed in the forums before. Instead of reviving a dead thread I started this one.

I've been comparing my car's fuel economy vs. other identical cars and noticed some inconsistencies in the way people enter data, which results in skewed averages (per tank, and overall).

For example this car: https://www.fuelly.com/car/honda/acc.../vancey/750667

A 2017 Honda Accord Hybrid can't travel more than ~700miles on a tank. Yet the car above has approx 15 fill-ups with mileage from ~950 to 1800. Clearly that's impossible. I'm sure it was a simple error on data entry, perhaps forgot to check the "missed fuel-up" option.

My suggestion:
+Using your pool of data by vehicle model, or from published specs, do your best to (statistically) estimate the size of the fuel tank and maximum expected range per fuel-up.
+flag outlier data that is just not logical or physically possible. Exclude this data from the displayed vehicle overall average.
+perhaps have a confirmation pop-up (with suggestion how to fix the error) when someone clicks save on an outlier entry?

I'm not advocating deleting or changing anyone's data. I'm advocating correcting the displayed average for the vehicle so one can more easily see how a car performs, and thus if your car is typical of the breed.

[Draigflag]

Yea I've noticed whilst flicking through and scanning other people's data, just the amount of incorrect/inaccurate data is quite overwhelming. Takes less than 10 seconds to find an obvious outlier regardless of make/model of car. I just use fuelly for my own data collection now.

[JockoT]

I already employ a filter. I look at the data and think, "That's $h*te".