Erratic MPG

[ironeagleuk]

Why is my MPG graph so erratic?

Is it because there are so many partial fill-ups as per your FAQs?

I have imported my data from another site.
Can you shed any light as to why entry 286 (for example) doesn't show an MPG?...on the other site I use it calculated 37.1mpg for that entry

[pb]

Yes, it's the partial fuel-ups. You might want to take a look at this FAQ in particular:

Why doesn't my fuel-up have a calculated MPG (L/100km)?

This gives several reasons why you might not get an individual MPG calculated for any single fuel-up. You need two full fuel-ups in a row to get an individual calculation for a fuel-up.

As for entry 286, you have an unusual fueling pattern that revealed a problem with our system. You enter a partial fuel-up, and then a full fuel-up on the same day. Because our system gave precedence to the date--and you don't specify the time in your fuel-ups--it wasn't sorting the fuel-ups properly for calculation. I fixed that up and you should be set going forward.

But yeah, the lack of calculated fuel-ups is largely due to the large number of partial fuel-ups you have. You might also take a look at this FAQ entry:

Why do I have to fill the tank all the way up every time I buy fuel?

[ironeagleuk]

Thanks for that.

I did study the FAQ's, and was just confused why there were blanks, but I hadn't noticed that the problems were fuellings on the same day.

I usually do partial fuelings from a local expensive garage when low and fill up just enough to get me to the cheaper petrol stations.

[campisi]

The way I deal with partial fill-ups is to combine the partial with my next full fill-up and use the miles driven total and gallons pumped total from BOTH fillups into one quasi-fillup. That way at least you use all of your data in calculating MPG. It has it's plus's and minus's but it gives perfectly valid data.

[MMUK]

I did the same, manually combined the partial with the next full tank to create a single supersized tank. It's unfortunate that fuelly can't do the same, but Paul wants to keep it as simple as possible.

I only had one partial and that was when I wasn't allowed to fill the tank myself.

[MMUK]

Of course if there are two partials in a row, that doesn't work (you have to add all consecutive partials together with the following full tank).

So in the example:

Tank A full

Tank B partial

Tank C partial

Tank D full

You get the mpg from the first tank (A) without problems. The next mpg figure comes from adding B, C and D together.

[pb]

Yeah, you can do what works for you. I think this method of combining partials won't give you an accurate number because the size of your tank won't be consistant. You don't get an accurate "difference from full".

We've been over this a lot--across dozens of conversations here in the forum--so I'm not really interested in having this conversation again. There are absolutely many ways you can calculate your fuel economy, but I don't feel like this method is accurate. My partner at Fuelly, Matt, feels the same way. We have talked about this many, many times. If we felt this method of combining partials gave you an accurate MPG we would implement it, but we don't think it does.

I don't want to disuade you from doing it on your own if you feel like that number you get is reliable, but we don't want to build it into the system. We'd rather have a blank space in the list of fuel-ups than misleading calculations. We use total distance driven and total fuel filled for the overall average numbers, and that takes partials into account. So all data is being used for your overall averages, including partials.

[gpk]

Hmmm newcomer here.... I was hoping I could use fuelly to cut out my spreadsheet work but it's a bit of a spanner in the works for me that fuelly doesn't show (using MMUK's example) the average MPG over fill-ups B, C and D. It's especially painful given how good the other features are!

Since this is such a frequent request and generator of so much heat (!) I wonder how it could be done so as to get round pb's concerns. It is certainly possible (easy) to calculate the average over the distance covered by fill-ups B, C and D and this is accurate just as the overall average figures that fuelly generates are as accurate as the combined fuel volume and mileage data supplied. And even the MPG figure for a single fill-up is only an average over the period covered by that fill-up - first half tank could be city, 2nd half highway but only a single average MPG can be calculated. It might even make things like the graph of MPG over time more accurate since for people who do some partial fills there will be gaps in the graph which will be interpolated over in a potentially misleading way. For example in the case of pb's Prius http://www.fuelly.com/driver/pb/prius if the fill-ups of Aug 23 and Sep 13 had each been done as a partial and then a top-up to full, those data points would be missing and you might mistakenly get the impression that economy had taken a nosedive in the fall. Whereas in all probability what has happened is that on Aug 23 the pump clicked off before full, and this is compensated for by the Aug 30 fill-up.

Anyway I wasn't going to re-open the old conversation ...honest......!! I appreciate that doing any more than what there is at present might require quite a bit of re-engineering to get everything into an acceptable form. Just logging a +1 for something like this as and when the time is right.

And I know it's impolite to ask here, but if anyone knows of a similar service that will show the averages like in MMUK's example........!!

[MMUK]

>And I know it's impolite to ask here, but if anyone knows of a similar service that

>will show the averages like in MMUK's example

As you have already discovered, you can export the tanks into a spreadsheet. You can do pretty much anything with it there, including calculating groups of tanks etc.

In my own spreadsheet, I don't try to handle partials directly (since I only have one), but I do have a chart which has a 3-point moving average, and that gives a good indication of what the underlying MPG is regardless of short-fills etc.

[gpk]

Thanks MMUK, I was hoping to get away from my spreadsheet and just do it all via updates from my phone before leaving the petrol station. Hence the search for a web-based solution...........!

Have a good w/e.... and BTW your MPG is pretty astonishing, especially as we seem to have the same engine. Mine seems to have dropped a few MPG over the past year (not visible from the data I've entered here) - wondering if Europarts QX oil has something to do with it...!