Fuelly MPG Going Up!!!

[fotomoto]

Two days later: 39.9

 

40.0 by the weekend?  :happy feet:

[RachelnLa]

Up up up and away!!!!

[fotomoto]

I did some digging to find out why or how there could be a driver with a 29mpg rating.  I found it's an owner with a gas car in the Netherlands.  <face palm>

 

More digging found another one with a 19.1mpg avg.  NiNETEEN!  I contacted him about the discrepancy (last entry was also over 6 months old) and he gladly removed all the data.  

[salsaguy]

+3, I agree there are too many variables to track and neither avg or median will accurately help to represent the expected FE of any  vehicle. but in median at least the few bogus and extreme values don't affect the data (group average mpg) as much as if you use avg. there is no answer. yes you can do a design of experiments to see which factors are most important to FE, like tires, weather, driver, % City driving, etc. I'm sure someone has already done this in the past. just admit that avg does have inherent problems.

Edited August 26, 2013 by salsaguy

[Adair]

• 2013

•  

 

 

40.2

Avg MPG

• 239

  Cars

• 3,609

  Fuel-Ups

1,437,275

Miles Tracked

VIEW ALL

 

[Fits2at]

My last 3 tanks of gas have all been over the average 47mpg with the highest being 52 mpg. I got 611 miles on that tank of gas.  We have been having an unusually cool summer so I haven't had to run the A/C. I haven't even been to get the update yet because I wanted to see what (if any) problems came from the update.  I love my Maxie!!

[tehuti]

I jus got my second fuelup, first stats for Fuelly. I am currently getting 37.3 mpg @ 855.5 miles. I was a bit disappointed that I didn't go longer than 7 days before I had to fuel up, but that will improve over time, if what I have been reading is true.

[fotomoto]

 

• 2013
•  
•  
•  
• 40.2

  Avg MPG

• •  

 

Actually dropped  .1 back down to 40.1   Hmmm.  Maybe a reflection of summer highway traveling?  I know I contributed with numerous sub 40 tanks.

Edited August 26, 2013 by fotomoto

[Andy from Fuelly]

Hey there, I just wanted to let you know that we're listening and plan to incorporate your ideas into a future release of Fuelly.  Right now we are working on a new foundation that will allow us to iterate faster in the future.  We're also working on a better next gen version of apps (iOS and Android) as we feel it's important to have a native solution which is as fast as possible to log your fuel-up at the pump (with an offline mode for syncing when a connection is available).  The apps will be connected to the new back-end foundation.

 

We are planning a better way to organize vehicles by using VINs and/or drop downs with more granular fields based on a reverse VIN type query where we can ask owners which options are on those vehicles as well.  As this all comes together we'll have the data to allow users to drill down with more granularity.

 

When it comes to Median vs Average, we'll probably allow users to filter and see both.  We'll also look at ways to analyze stats based on all fuel-ups for a specific search, not average for a single car and then average for that specific make/model/options.

 

We'll be setting up an Ideas section soon so you can add your ideas and vote them up.  I'll also stop by here to read your Fuelly related posts so feel free to post your ideas here, I'll read them.

 

Thanks for being Fuelly members, it's amazing what we can all do together!

[salsaguy]

you can't just look at the 2013 mode.l you have to remove the non hybrids from that sample as they are not the same car or FE numbers.  it doesn't show you a graph sadly when you do use the filter to get rid of those. you have to manually calculate by adding up all the remaining ones which is a hassle.

 

it IS like the housing market as FE changes based on location, weather, tire inflation, driver style, change in elevation during commute, city driver vs highway driver, A/C usage, hypermiling or not, updated model with FE software patch or not, etc etc etc....

 

and there IS a difference between the numbers between median vs Avg. go ahead and run the numbers yourself to see. it's not a staggering number but when you guys are up on the 45-65 mpg ranges a little change/few 0.5 mp IS a big deal.

 

Average is not worthless in fuelly because it's based on total miles driven divided by gallons consumed.   

 

This is not the housing market where the sales value varies by square footage, location, amenities and so forth such that one $10 million sale might be the equivalent of 50 median priced home sales.  In the housing market, median is appropriate.  In FE (assuming sample size is appropriate), average or median is fine as there's likely little difference between the two.

Edited August 26, 2013 by salsaguy

[salsaguy]

very very cool Andy. love to hear you are planning on improvement. how did you find us here by the way???

one thing for sure I'd like to see is after you filter you results it will show you the updated graph with the updated data set you have drilled down to get to so you don't have to manually calculate the info once filtered.

keep up the great work. like the VIN verification idea to keep the bogus leeches who try and change the avg due to an ulterior motive on their part.

 

Hey there, I just wanted to let you know that we're listening and plan to incorporate your ideas into a future release of Fuelly.  Right now we are working on a new foundation that will allow us to iterate faster in the future.  We're also working on a better next gen version of apps (iOS and Android) as we feel it's important to have a native solution which is as fast as possible to log your fuel-up at the pump (with an offline mode for syncing when a connection is available).  The apps will be connected to the new back-end foundation.

 

We are planning a better way to organize vehicles by using VINs and/or drop downs with more granular fields based on a reverse VIN type query where we can ask owners which options are on those vehicles as well.  As this all comes together we'll have the data to allow users to drill down with more granularity.

 

When it comes to Median vs Average, we'll probably allow users to filter and see both.  We'll also look at ways to analyze stats based on all fuel-ups for a specific search, not average for a single car and then average for that specific make/model/options.

 

We'll be setting up an Ideas section soon so you can add your ideas and vote them up.  I'll also stop by here to read your Fuelly related posts so feel free to post your ideas here, I'll read them.

 

Thanks for being Fuelly members, it's amazing what we can all do together!

[Adair]

Hey there, I just wanted to let you know that we're listening and plan to incorporate your ideas into a future release of Fuelly.  Right now we are working on a new foundation that will allow us to iterate faster in the future.  We're also working on a better next gen version of apps (iOS and Android) as we feel it's important to have a native solution which is as fast as possible to log your fuel-up at the pump (with an offline mode for syncing when a connection is available).  The apps will be connected to the new back-end foundation.

 

We are planning a better way to organize vehicles by using VINs and/or drop downs with more granular fields based on a reverse VIN type query where we can ask owners which options are on those vehicles as well.  As this all comes together we'll have the data to allow users to drill down with more granularity.

 

When it comes to Median vs Average, we'll probably allow users to filter and see both.  We'll also look at ways to analyze stats based on all fuel-ups for a specific search, not average for a single car and then average for that specific make/model/options.

 

We'll be setting up an Ideas section soon so you can add your ideas and vote them up.  I'll also stop by here to read your Fuelly related posts so feel free to post your ideas here, I'll read them.

 

Thanks for being Fuelly members, it's amazing what we can all do together!

Hi Andy, and thanks for your information.

 

Since hybrids are so impacted by climate, I think a seasonal filter would be helpful......and because *winter* means different things to different areas....... to be able to select the group of months you want to check......*winter* in Minnesota lasts longer and is harsher than winter in other areas. Just a thought.

[Testdriver]

Hey there, I just wanted to let you know that we're listening and plan to incorporate your ideas into a future release of Fuelly.  Right now we are working on a new foundation that will allow us to iterate faster in the future.  We're also working on a better next gen version of apps (iOS and Android) as we feel it's important to have a native solution which is as fast as possible to log your fuel-up at the pump (with an offline mode for syncing when a connection is available).  The apps will be connected to the new back-end foundation.

 

We are planning a better way to organize vehicles by using VINs and/or drop downs with more granular fields based on a reverse VIN type query where we can ask owners which options are on those vehicles as well.  As this all comes together we'll have the data to allow users to drill down with more granularity.

 

When it comes to Median vs Average, we'll probably allow users to filter and see both.  We'll also look at ways to analyze stats based on all fuel-ups for a specific search, not average for a single car and then average for that specific make/model/options.

 

We'll be setting up an Ideas section soon so you can add your ideas and vote them up.  I'll also stop by here to read your Fuelly related posts so feel free to post your ideas here, I'll read them.

 

Thanks for being Fuelly members, it's amazing what we can all do together!

 

Does this "new foundation" include a RESTful API that will be available to developers? A couple months ago I sifted through the Fuelly client code and noticed that it was possible to pull most of the Fuelly fill-up data in raw JSON format, but it would be kinda a pain in the butt and I never bothered with it. A real API for developers would be nice.

[Andy from Fuelly]

very very cool Andy. love to hear you are planning on improvement. how did you find us here by the way???

one thing for sure I'd like to see is after you filter you results it will show you the updated graph with the updated data set you have drilled down to get to so you don't have to manually calculate the info once filtered.

keep up the great work. like the VIN verification idea to keep the bogus leeches who try and change the avg due to an ulterior motive on their part.

 

I found you doing a search for Fuelly mentions in recent weeks.  I'm planning to try and reach out to our members on their "home forum".  I am a forum guy through-and-through and know how often I turn to my forum buddies to discuss something.  My forum friends often point me in the right direction pretty quickly.  So my thinking is that Fuelly needs to go where our users are, read what people are saying, ask questions and learn how we can make our product better.

 

I agree about the charts and data, it should show you numbers for how ever granular you want to filter things.  I am hoping we can also define some other interesting stats from our data like distance per tank, average time between fuel-ups, Average cost/mile for driving (once again you could filter by locations such as state, southern states, northern states, etc. View averages by month, etc and mixes of views so you could view performance of a hybrid in northern states (ie. colder winters) by month and view seasonality in the charts.  The idea is that we have all this data, lets try to find some interesting ways to show the data so everyone can learn from each other.

 

As for bogus leeches, we'll work on some algorithms to find them.  My background is with online forums (like this one) and we've found some pretty effective ways to find shills in the forums and will use some of those techniques as well.  Our goal will be to develop an algorithm that understands a patterns for a specific make/model and can alert us to a potentially bogus user.  I think we'll get pretty good at spotting them and removing their data from our calculations...

 

Hi Andy, and thanks for your information.

 

Since hybrids are so impacted by climate, I think a seasonal filter would be helpful......and because *winter* means different things to different areas....... to be able to select the group of months you want to check......*winter* in Minnesota lasts longer and is harsher than winter in other areas. Just a thought.

 

I agree about seasonality and geographic location.  We are planning to handle this data by collection location of fill-ups from mobiles.  For example if fuel-up #16 was in Minneapolis and fuel-up #17 was also in Minneapolis and the time between fuel-ups looks "normal" then we can assume you have been driving in your home town of Minneapolis.  From there we can request local weather info and append the actual temps in your area for the days between Fuel-Ups.  We'll then have a profile where you can tell the system when you normally drive and it will use this data to determine an average temp for your fuel-up.  We'll constantly look for better ways to gather accurate data, for example, once we have an ODBII reader it will know the exact times you drive and we'll get even more accurate temps.  Temp is important for any batter powered vehicle.

 

Does this "new foundation" include a RESTful API that will be available to developers? A couple months ago I sifted through the Fuelly client code and noticed that it was possible to pull most of the Fuelly fill-up data in raw JSON format, but it would be kinda a pain in the butt and I never bothered with it. A real API for developers would be nice.

 

Yes, that's exactly what the "new foundation" is going to be.  We'll be rolling this API out internally and then to the public.  The data will probably be provided under the Creative Commons Attribution Required (non-commercial) license with some type of rate controls for new accounts and then unlock the volume once we understand how someone is going to use the data.  We want the data to be open and available but we also want to protect it (for example, unethical uses, etc).  We'll have to find the right balance.

Edited August 27, 2013 by Andy from Fuelly

[salsaguy]

very very cool stuff Andy.so glad you are reaching out and learning from the " customer"

in any business that's the key..to know your customers needs, wants and try to not only meet the needs but excel and do even better so as to delight them not most net the minimum requirements (ie the kano model...look it up) .... so do i sound like a quality engineer yet??? (maybe because i am) :)

 

so while we're at it...can we please be able to export the Fuelly data into excel, csv or even better for Minitab statistucal software use (yes i know Minitab can import excel files/data) so that we can manipulate it ourselves to do our own analyses?

Also a better way to separate the models especially if the mfg offers both a gas version and a hybrid version so we don't end up with what we had happen with the cmax ...with some users falsely reporting their car as a hybrid when it clearly wasn't (European models which are gas only) for the 2013 model.

 

thank you again for all the help and listening to the crowd.

[Andy from Fuelly]

You can already export your data.  Here is the FAQ for that:

http://www.fuelly.com/faq/9/Can-I-export-my-fuelup-data