RunKeeper inconsistent results?

I'm using the RunKeeper app to track power walks and love how it automatically updates MFP with my calories burned. Today I did a test to see how consistent it is. I walked one way for awhile while it tracked me, then stopped and saved the reported metrics. I started a new walk and let it track me back along the same trail and I walked at the same speed.

One walk estimated 125 calories, the return walk estimated 99. I know these are estimates and depend upon several variables, including the quality of the GPS signal. But I followed the exact same route so would have expected far more similar results. A 25% deviation is a bit high in my option.

htzS4kZ.jpg
k6y3P7f.jpg

Maybe I am using it wrong?



«1

Replies

  • yesimpson
    yesimpson Posts: 1,372 Member
    Probably a stupid question, but was the trail uphill one way and downhill on the other leg?
    I do notice some inaccuracies with RunKeeper but couldn't tell you if it is better or worse than any other. It's always going to be an estimate at the end of the day, so as long as you take it with a pinch of salt you'll be OK!
  • tigerblue
    tigerblue Posts: 1,526 Member
    I have used "Log Your Run" for awhile now. I find it to be very inaccurate. When I turn corners, it shows that I slow down from like 10 minute miles to 13 minute miles. It really only seems accurate if I am travelling directly north-south or directly east-west. And if I go to the website and track my route manually there is significant difference in the distance.

    I'm guessing that both the app that OP is talking about and my app rely on the location services of our smartphones. They both seem inaccurate. Mine won't report directly to MFP, so I rely on a Misfit activity tracker and the MFP estimates instead.

    Bottom line, I am interested in what others have to say about this. And I would love to know what the best run/walk tracking smartphone app out there is!
  • MeanderingMammal
    MeanderingMammal Posts: 7,866 Member
    Your distance is so short that the level of error inherent in using phone based GPS is exacerbated. That deviation could be accounted or by the changes in position that chopping between spacecraft as they cross the horizon.

    That said, it's still overestimating your calorie expenditure, I'd anticipate about 60-70 cals at most.
  • W_Stewart
    W_Stewart Posts: 237 Member
    That said, it's still overestimating your calorie expenditure, I'd anticipate about 60-70 cals at most.

    What can I compare my calories burned output to? I should be able to enter the distance/speed.

  • W_Stewart
    W_Stewart Posts: 237 Member
    yesimpson wrote: »
    Probably a stupid question, but was the trail uphill one way and downhill on the other leg?
    !

    I don't think it can track elevation.

  • MeanderingMammal
    MeanderingMammal Posts: 7,866 Member
    wahoowad wrote: »
    What can I compare my calories burned output to? I should be able to enter the distance/speed.

    Just use MFP
  • yesimpson
    yesimpson Posts: 1,372 Member

    wahoowad wrote: »
    yesimpson wrote: »
    Probably a stupid question, but was the trail uphill one way and downhill on the other leg?
    !

    I don't think it can track elevation.

    It does. I'm assuming you have an iPhone - if you tap on the Me button in the bottom panel at the far left, it opens up a screen which displayed all your tracked Activities. Click on any Activity, scroll down to Charts and tap that - it does roughly gauge your elevation, which will affect its calorie estimates for exercise. Whether that's what happened with the example above I don't know, but from using it I think I have definitely been given higher estimates for hilly trail runs than flat pavement ones.
  • shroodle88
    shroodle88 Posts: 123 Member
    I've had similar issues with tracking my 13km ride to work. Sometimes I get differences of ca. 500kj even though the time and route stay pretty much the same. If in doubt, go with the lower estimate - you might be in for a nice surprise later. :-)
  • yirara
    yirara Posts: 9,838 Member
    It does track elevation, and based on my observation it overestimates elevation for quite a bit. I live in a completely flat area and still got 300m elevation gain over 5km runs. There's a solution I found on their website, though it's a bit fiddly: go to the workout you logged. Right top of the map is a triangle. Use that and chose the Edit option. On the map you have to uncheck 'stick to roads' at the top and then zoom in so you can see one data point clearly. Touch it and only move it the smallest possible distance. Make sure you're not creating additional data points (distance, pace, etc will change). Save it only when you're happy. The elevation will be more realistic and the calorie burn as well.
  • yirara
    yirara Posts: 9,838 Member
    Btw, since I'm using a different app to log my runs (it's sends data to the runkeeper website) I'm getting more realistic numbers. The problem must be somewhere at the interaction between their own app and their website.
  • tigerblue
    tigerblue Posts: 1,526 Member
    yirara wrote: »
    Btw, since I'm using a different app to log my runs (it's sends data to the runkeeper website) I'm getting more realistic numbers. The problem must be somewhere at the interaction between their own app and their website.

    What app are you using?

  • yirara
    yirara Posts: 9,838 Member
    It's a paid app for iOS only called iSmoothRun. It's got some really nice stats which you can view in-app. It's certainly more nerdy and has a lot of custom settings, which I like. I send my running data to RunKeeper and another running site called Smashrun. If I keep up running and the annoying health issue related to running that bugged me in the past doesn't come back I'm considering becoming paid member there. But then again I do love stats.
  • pcrucifer
    pcrucifer Posts: 71 Member
    Most apps are going to overestimate distance and underestimate speed. I run in a rural Midwestern US setting where all the roads are set on a 1 mile grid, so it is easy to compare the apps to known distance. RunKeeper is pretty good, usually less than 1% on a several mile run. However, it is inconsistent on elevation. I generally assume that RK is an overly encouraging friend who always tells me I went faster and burned more calories than I did. Not lying, but definitely seeing the world through rose colored glasses.
  • yirara
    yirara Posts: 9,838 Member
    edited April 2015
    I think the inconsistancy with distance is due to a multitude of different things. There's sampling rate. Most apps try to keep the data volume down and hence create relatively little data points, others resample your data to reduce file size. And a large part of inaccuracy is due to how the distance is calculated. Running along a straight road is not simply the direct distance between A and B, but the curvature of the Earth needs to be taken into account. If altitude is added in that then the equation gets seriously complex. Different apps have different solutions for that. This might be a good websource to get you started on it: http://trac.osgeo.org/proj/wiki/GeodesicCalculations#GeodesicCalculations

    Btw, next to runkeeper I'm mainly using this stats site now. It's pretty cool for the nerdy-minded folks: http://smashrun.com/yirara.ra/invite add me if you like.

  • W_Stewart
    W_Stewart Posts: 237 Member
    yesimpson wrote: »
    It does. I'm assuming you have an iPhone - if you tap on the Me button in the bottom panel at the far left, it opens up a screen which displayed all your tracked Activities. Click on any Activity, scroll down to Charts and tap that - it does roughly gauge your elevation, which will affect its calorie estimates for exercise. Whether that's what happened with the example above I don't know, but from using it I think I have definitely been given higher estimates for hilly trail runs than flat pavement ones.

    Ahh. I see that now. It is hilly here and I am glad it takes that into account for my calorie burn. That said, I'm getting increasingly skeptical of some of the calorie burn figures it is producing. I did what someone suggested and compared it to directly entering the distance into MFP and MFP is frequently lower by around 25%. I guess I'll try it on some flat walks and see what I get.

  • MeanderingMammal
    MeanderingMammal Posts: 7,866 Member
    Accuracy around elevation is driven by the number of GPS channels available. Four spacecraft is the minimum required for a fix, and most phones are limited to four channels. Elevation comes from having more spacecraft in sight, with the average visibility being nine veicles, with up to twelve being possible. Most higher end dedicated devices can cope with between 9 and 12 channels.

    So with a phone, the elevation is always going to be a bit less accurate in terms of direct measurement, so the various applications depend on alignment with known mapping. Some are better than others.
  • yesimpson
    yesimpson Posts: 1,372 Member
    edited April 2015
    wahoowad wrote: »
    yesimpson wrote: »
    It does. I'm assuming you have an iPhone - if you tap on the Me button in the bottom panel at the far left, it opens up a screen which displayed all your tracked Activities. Click on any Activity, scroll down to Charts and tap that - it does roughly gauge your elevation, which will affect its calorie estimates for exercise. Whether that's what happened with the example above I don't know, but from using it I think I have definitely been given higher estimates for hilly trail runs than flat pavement ones.

    Ahh. I see that now. It is hilly here and I am glad it takes that into account for my calorie burn. That said, I'm getting increasingly skeptical of some of the calorie burn figures it is producing. I did what someone suggested and compared it to directly entering the distance into MFP and MFP is frequently lower by around 25%. I guess I'll try it on some flat walks and see what I get.

    That's interesting, because I find it reports about 40% lower for me than MFP, and often the calorie estimate will be inflated when the data is pulled through to MFP, so I take Runkeeper as more realistic. So a run I was thinking would add 300 extra calories to my food diary often appears as 420+ on MFP, so I have to adjust it back manually. Funny how different individual results can be - I haven't gained any weight eating back every calorie Runkeeper says I've burnt. I think perhaps I am the (lucky) odd one out.
  • W_Stewart
    W_Stewart Posts: 237 Member
    yirara wrote: »
    There's a solution I found on their website, though it's a bit fiddly: go to the workout you logged. Right top of the map is a triangle. Use that and chose the Edit option. On the map you have to uncheck 'stick to roads' at the top and then zoom in so you can see one data point clearly. Touch it and only move it the smallest possible distance.

    I just tried this and nothing changed....? I only moved the data point a smidge, but it moved. Then I saved it and nothing changed. Does this only apply to changes to elevation?

    Hmmm...I did it again and this time it updated. Can you post a link to this solution? I'd like to read up on how they explain what it is doing differently. The calories did jump down though, and when I manually entered the same walking distance into MFP the calories burned were much, much closer.

  • yirara
    yirara Posts: 9,838 Member
    Well.. there's some information in their help files, but it's not really clear. It looks like the data gets processed via an online elevation model. I checked the model for my usual running grounds and the elevation is like expected but as soon as the runkeeper data moves through this thing and gets back to their website/app the elevation data is messed up.

    There is some information here: https://support.runkeeper.com/hc/en-us/articles/201998609-Elevation-Calorie-Issues
    and here: https://support.runkeeper.com/hc/en-us/articles/201109736-How-does-RunKeeper-calculate-elevation-and-climb-

    While they say their elevation data might sometimes be off it's so far always been seriously off for me. I live in a very flat area. If I'm seeing some elevation gain of over 300m over 6km then it's just wrong. I do think the 50something I get after editing are much more realistic.
  • W_Stewart
    W_Stewart Posts: 237 Member
    After editing are you talking about the elevation being updated or the calorie burn being updated?