suggestion: running averages of data?
 

Just connecting the dots of MPG data is a bit misleading.

It would be nice if there was some running average curve instead of a connect the dot curve. Post a smooth curve, like posting a point at each fill up containing the average of the last 3 or 5 fill ups. That will smooth out variations on each fill due to difference in tank fill level, or mileage readings. Look at Loess curves for example.

It would also be nice if the X axis of the plot was the date, not the fill up number. Then you'd see gas usage over time.

While I'm at it. A plot of mileage/$ or gas price over time would be a neat add also.

good idea!

another way of implementing it would be a trend line with shaded areas that show statistical confidence:

Link button did not work, I will just post it as text:

https://folding.extremeoverclocking.c...ry.php?s=&t=32

1. Yeah, a running average would be nice. We have some pretty limited graphing toolkits currently, so I'll have to look at what is possible at the moment.

2. We don't do the date since people on roadtrips can have more than one fill up on a single day (it happened often and was reported as a bug), so we had to instead change it to every fuel up gets a spot on the graph.

3. We're hesitant to include price in too many metrics because it's not something you have tons of control over. We're trying to highlight the things you can change with behavior -- using a lighter foot will save you money and you'll get farther, but with the fluctuations in fuel prices, we don't want to highlight too many things based on price because the price is almost completely out of driver's control.

If you saved $50 this month, it's not because you changed driving habits for the better, it's because gas is 40 cents less per gallon, etc. and there's no way to predict that going forward.

For #2, I think that you can plot this on a scatter plot instead of x-y plot to allow multiple fills on the same day. You could have an option to show either way also.