Fitbit App distance estimation

[SunsetRunner]

I'll start by saying that I'm not yet using a band (waiting for charge 3 shipping), but I already started to use Fitbit app for get familiarity with that. I'm noticing that distance is overestimated even if stride length auto-calculated is almost right (71.2 cm). Without spending too much by writing let's go to datas:

today I go hiking and used Komoot for path, it says about 21 km, this is what i get from Fitbit app compared to S Health:

Fitbit App:

steps: 34 729

distance: 41.04

cals: 3955k

S Health:

steps: 34 761

distance: 21.74

cals: 2577k

PS. I noticed that every time I open the app it shows like doubled steps until it gets updated, for example today opening it, it shows 64k steps and after 10 secs it gets corrected to right value

[Rich_Laue]

A couple of thoughts I have @SunsetRunner

Is the Fitbit app up to date?

You say that you where out for a walk, did you use the mobile Walk mode with the GPS? Do you see a map of your walk?

Normally Fitbit uses the stride for distance calculation, but if the app was using GPS, then the distance should be measured.

[SunsetRunner]

Hi @Rich_Laue, thanks you for your reply

App is in its last version on Google Play.

I didn't tracked the walk with GPS, but sometimes I opened maps to see the route. In Fitbit app I don't see any kind of map, so my thought is that the app should use stride for calculation.

Now, I don't know if it uses an algorithm for a proportion between walk stride and running stride based on cadence. Doing some rough calculus:

with walk stride: 34.729 * 0,712 = 24,727 km

with running stride: 34.729 * 1,12 = 38,896 km

that is still less than 41 km

Note: this problem happened even before, when I walk in city but I thought that the error was due to my fast walking speed that misled walking or running right detection so I did't really cared about

[Rich_Laue]

Then I'm not sure how you tracked your walk @SunsetRunner, did you track it on the wrist, or the Fitbit app. The Fitbit app should have given you a map of the walk.

If on the wrist, and the phone nearby, you could have used the connected GPS and received a map.

However without a map tgen Fitbit can only take an educated guess at the distance based on the determined stride.

It is more accurate to manually set the stride settings.

Yes Fitbit will try and calculate the stride based on age, height, gender, etc, but it assumes the user is an average person. Most likely they are not.

As for running yes it has an algorithm to determine if the user is walking or running, then uses the appropriate setting. It does not make any adjustments for any changes in the speed. 

[SunsetRunner]

I didn't tracked anything just walk during the day.

Actually I'm just using the app, with MobileTrack for steps counting. The problem is that the distance based on stride looks awfully wrong, while steps and stride looks right.

Yesterday as I woke up I tried to manually set walking stride to 62.5 cm (using S Health datas from the previous message) into Fitbit app and resetting running stride.

At the end of the day the summaries from S Health and Fitbit looks similar about steps (S: 19 679 F: 19 664) but while Samsung estimates distance 15.28 km (77.6 cm ave stride) Fitbit app claims 22.91 km (116.5 cm). Now looking in Fitbit settings under stride length it says: walking 62.5 cm (as I set) and running 119.7 cm (estimated by Fitbit) so I suppose it calculates distance as 1098 steps walking and 18566 running (1098*0.625+18566*1.197=22.909 km). At this point I think the only thing I should try is to cancel my order for Charge 3 as probably the algorithm for distance estimation works is similar to their app, am I right?

[Rich_Laue]

Throughout the day, how often does the average person actually walking with their normal out for a walk stride? If you think about it, not very often.

For this reason I'm not sure if it is even possible to calculate a reasonable guess of the distance let's say while working in the kitchen. Do you always 

Also comparing a device that like Fitbit that looks at arm motions, to a device that looks at the bounce of the  body core will probably not match up every day in step for not. However I did have a few months when their monthly totals where within 100 steps.

When neither S-Healrh or Fitbit claim to measure distance, with both claiming to count steps, I'm trying to understand why one would not start off by first compairing what they both claim to count rather then start off by trying to compare what they both claim to take a guess at. If the 2 devices cant agree with what they count, then how can they agree in what they take a guess at. It sounds like your putting the proverbial cart before the horse. 

Just saying that you first need to confirm what they say that they do count, rather than trying to compare what both say that they can not count. 

[SunsetRunner]

I'm totally conscious that the way an arm fit tracker works is completely different from the way a software counts steps based by movement of a third-party device in my pocket. In fact I'm not against steps counting that I consider enough accurate (I don't care if Fitbit and S-Health have 500 steps a day of gap).

I was just complaining about distance estimation (agree that couldn't be right): if I were a software engineer I wouldn't implement a distance estimation if it's so inaccurate. For example, at this moment of today I just grabbed my phone around house and both apps says I did about 850 steps but while S-Health guess about 0.62 km (maybe totally wrong but it looks possible value) Fitbit guess 0.87 km that means steps of 1+ meter (or maybe some steps in kitchen of 40 cm and some other of 2 meters moving from kitchen to bedroom): I hope you agree how absurd this value is.

I'm going to buy Fitbit band not for get an accurate distance (as I didn't buy my phone for having it) but for example to use it going run without phone and if I do 10 km on a athletic track I hope it guess a distance between 8.5 km and 11.5 km (even 20% more or less could be nice, but guessing 200+% I think something is wrong)

My intention is to understand if I'll get this type of deviation going to buy a Fitbit: could you tell me about some of distance that yours devices try to estimate?

Thanks you!

[Rich_Laue · Accepted solution]

Over a two week period, I did 10 walks on ten different days over the same course.

I then added together the steps recorded for each walk and divided by (10*the distance per walk).

This gave me an average stride based on a two week period. Yes the stride will change based on both the physical and mental pains and problems.

I then set my Fitbit stride to match my average stride

I now find the reported distance during walks to be fairly accurate.

As for around the house, I do not expect Fitbit to get this correct especially do to constraints my stride is no where close to that of my walks.

Not knowing how S-Health works, could they have multiple stride settings for walking? One for when steps are not consistent as when working around the house, and another for when out on a walk? 

[SunsetRunner]

I'll do so, and manually set stride length consequently. If I remember right from Charge 2 there should be a features for auto-calibrate the stride when walks and runs are GPS tracked.

Thanks you for your time!

[Rich_Laue]

You may want to read the user manual again, the Charge 2 has an auto stride setting but this simply uses the users age, height, gender etc to calculate an average stride for a person with these stats.

The Ionic and maybe the Versa are the only Fitbits that use GPS during a run to auto calcalculate the stride. 

[SunsetRunner]

Here's where I read that: https://help.fitbit.com/articles/en_US/Help_article/1135

So I suppose even Charge 3 will have that same feature...