Reasonability checks on input

[Cookie]

Do you have checks on the reasonability of data entered? Just after entering a fill up for my Civic that calculated to the excellent 6.0 L/100km I noticed a recent fill up for somebody else's Civic that was supposedly 2.7 L/100km (or about 48 mpg!). I assume it was because of a typo.

I imagine it would be difficult to set standards because of the wide range of vehicle types but perhaps a comparison to the historical record of the particular vehicle would work.

The vehicle in question is at http://www.fuelly.com/driver/barefootvet/civic

[mathowie]

Yeah, we have reasonability checks in place, but we frequently find people (esp. those in high mileage clubs) break the sanity checks we originally had in place. So we have some checks still around, but they're flexible enough for the rare person hypermiling a hybrid car to 70mpg, but should block someone accidentally entering in 1,000 gallons of fuel at a stop or that they drove 3,500 miles on a single tank.

[pb]

Yes, we do have reality checks in place when you enter data. That includes a maximum distance traveled and a maximum fuel volume. The exact numbers are different for motorcycles vs. cars, but we don't adjust the limits for specific models.

Fuelly is completely dependent on user input, and sometimes people simply enter bad data that falls within our data limits. In this particular case, Fuelly thinks that civic went several hundred miles on 4 or 5 gallons of gas. (It probably should have been marked as a partial fill.) Definitely an outlier, but we're hoping people have the incentive to clean up their data once they see it's completely unrealistic. We'd like to err on the side of being too inclusive rather than exclusive when it comes to entering data, because constantly bumping up against the limits when you're at the pump is frustrating.

[Cookie]

Thanks mathowie and pb.

Rather than block questionable input perhaps an "Are you sure?" routine if the mileage was (say) 20% better or worse than the average of previous fills?

[jhu]

48 mpg on a Civic is doable, especially if it's a manual since there are a few things you can do in a manual that you can't do with an automatic. Hell, I used to get ~40 mpg/tank on my '96 Civic, which was an automatic.

[jhu]

Hmm, actually 2.7 L/100km is 87 mpg. That's not really a practical possibility on a non-hybrid Civic, which would entail something like pulse & gliding between 10-20 mph with the engine off while coasting and doing that for the entire tank. However, on a hybrid Civic, it should be a practical possibility.

[Cookie]

You are quite right jhu. It was my turn for a typo. I actually used the "correct" 87 mpg figure in a message I sent to the owner of that Civic but somehow it became 48 in my forum post.

[pb]

Well, we already have a line at the bottom after a fill that says: "Something look off? You can edit this fuel-up." And "edit..." is a link to change things. So I'm not sure that blocking the entry first gets us much more.

I think we just need to continue to trust people, hope they edit bad data, and live with some wacky reporting.

[FordMan]

48 MPG should be very doable in a Civic with a manual transmission. My '88 Escort Pony has 494K+ miles on it, and by the odometer which is registering less miles than I'm actually putting on the car because of larger tires is averaging 43.7 MPG and using the mileage my GPS says I'm traveling the average mileage is at 46.2 MPG. If you'd like to see my gas logs the mileage according to the odometer is listed under the car name "Old Reliable" and the GPS corrected mileage is listed under the name "1988 Escort". I recently took that car on a short highway trip and according to the miles registered by my GPS I got 54.3 MPG on the highway. 45-47 MPG combined (according to the odometer) and 47-50 MPG (according to the GPS) on recent tanks has become normal mileage for it.

[mdrehanrizvi]

we should allow user to enter the data they want. As it is upto the user to fill correct data for his vehicles