Quote:
Originally Posted by Rdean58
Hi all,
I thought that I would share some knowledge with you. Here is the way process analysis works (which is what we are doing with the fuel mileage tracking).
In order to establish a "base line" there needs to be at least 25 observances (fillups). Depending on how far you drive this could take a year 
Once you establish your base line, then you can determine your "mean average" by taking all 25 fill ups, adding them together, then dividing by 25.
|
<snipped out more statistical stuff>
I'm no statistical guru, but wouldn't the most accurate MPG numbers be given by simply adding up all of the miles you drove by all of the gas you put in? That would give you your ACTUAL lifetime MPG's.
Your method has you averaging the averages, which will give you an idea of the variances in individual measurements, but gives LESS accurate info on actual MPG's over the life of the car. This difference can be significant, especially if you fill up inconsistantly, e.g. sometimes with a near empty tank, and sometimes with a 1/2 full tank.
For example, using some made up numbrers:
Fill up and drive in mixed traffic till the E light comes on.
300 miles using 9 gals = 33.33 - ok so far
Now fill up and get in city traffic, fill up again after 100 miles
100 miles using 4 gallons = 25mpg
Whats your average?
True average = (300 + 100) / (9+4) = 30.76
Average of averages = (33.33 + 25) /2 = 29.16
<quote>
If you get a milage figure that is outside the norm, (like for instance 40 mpg or 15 mpg) you need to analyze why that happened. Once you have determine why (high tail wind, 3 hour idling time, etc.) then this observation is removed from your running observations.</quote>
Agree here, but I'd just remove this one measurement from your total and still calculate your 'true average' just leaving out this one measurement.
Easy to do in Excel. Just do a line item for each tank, but then calculate a lifetime average based on total gallons/total miles rather than averaging averages....