I've started keeping an OO.o spreadsheet of my fueling activity.

It contains the date, miles on the trip meter, gallons purchased, the reported MPG for the tank from the computer, and the caluclated MPG from the miles/gallons. I then look at the error rate and I can figure out lifetime MPG as well as the equivalent number from the reported MPG (take the reported MPG, multiply by the miles driven, add all the tanks together then divide by the sum of the miles driven). It sounds complicated, but I really only have to enter in the miles, gallons and computer MPG and the rest is calculated automatically. The only problem I have is I am not fluent enough in OO.o yet to make it do any nifty charts or anything. Maybe if I get it a little more polished I'll offer it as a template if anyone wants to use it.

 

Yes, it's essentially the same thing. I was simply showing how without actually having a value for gallons, I'm still able to calculate lifetime mileage. It all has to do with which values are actually stored in the database, but in the end it's all gravy.

 

Steve,
Is that not the same thing that the database does?

 

The database does not have a mechanism to track the difference between what the car's computer reports for mileage and what I get when I divide miles/gallons on each fillup.

So far they are not the same, either. The computer has been off by as much as 1.5 MPG and as little as 0.57 MPG (6 tank average error is 0.76 MPG).

 

Many enthusiasts have found that over the course of tens of thousands of miles, the difference between manual calculations and the display is about 1 MPG. Since not everyone uses manual calculations -- in fact, very few -- it's safe to say that the database's figures are accurate to less than 1MPG. We can't be sure of the exact error since we don't know who uses which method. My guess is that it's much less than this.

 

Really? Many people in other forums are heading in the opposite direction. They have become mistrustful of the computer and are of the belief they can be sufficiently accurate in their fillup technique that their manual calculations are more accurate. Here in the land of no self-serve I'm not so sure of that, but I find averaging both over several tanks has been illuminating.

 

I was just reading up on a thread online in which multiple owners reported their error around 1MPG over tens of thousands of miles. Of course, we're all looking to be accurate, but I've got a question for you: quality statistics are only good if they're compared to quality statistics. Even if our figures are 100% on the nose accurate, how much do you want to bet that Average Joe's figures aren't? Unless Average Joe's stats are right on the button, too, comparisons between his car and the database would serve little purpose. Right?

Well, no. Mileage is an approximation. It's got to be. Because, no one is 100% dead certain that their stat is perfect. Anyway, to most anyone who bothers to compare, 52 MPG is the same as 51. 26 is the same as 27. This is reality.

 

Hi Stevewa:

___I cannot say what others are doing in other forums but if you cannot trust the odometer in your car and the gallons pumped from your local station, relying on a trip computer is a far cry from accuracy if that is what you are really after.

___Along a similar note, take a look at the attachment of my Insight’s lmpg. Now take a look at the RHMDB with the tank over tanks calculated using the actual miles traveled/actual gallons filled. Afterwards, do you see any discrepancy?

Little Red Beauty

___I will never use the FCD for an mpg calc but at least it tells you ~ what you received over a segment, tank, lifetime, whatever. It just so happens my Insight’s lmpg matches my actual to a tee for whatever reason?

___Jason’s RHMDB is a good overall view of the mileage received in any number of automobiles and with the rounding, the over reporting, under reporting, and incorrect reporting, I bet the list is within .5 mpg of actual when looking over a type of hybrid or non-hybrid. No guarantees of course but I can bet it is very very close.

___Good Luck

___Wayne R. Gerdes
___Hunt Club Farms Landscaping Ltd.
___Waynegerdes@earthlink.net

 

Right. These are not controlled environments. There are no checks for user error. Even if I "required" members to input distance and gallons, there's no guarantee that they'd do it right.

I am confident that our figures are accurate for all practical purposes. For scientific purposes? No. But, for the sake of comparison, definitely.

 

Ah. Also another important thing to point out: notice the database uses medians instead of means by default. That means that when calculating mileage for a model, every single car's data is thrown out except for one (or two in case of an even number of vehicles).