In the Fitbit data you'll always see lots of zeros for steps. This could be because the user is not moving or because they are not wearing the Fitbit. For my application it would be useful to distinguish these two things. I was wondering, is there any way to determine it directly? In theory it should be possible, since even though you're not moving, if you're wearing the wristband there will be much more variation in the accelerometer data it captures than if it was stationary on a table somewhere, or even if it was dead.
But the API doesn't seem to give you access to any information like that.
Right now the best estimate I can think of is checking whether the gap (of minutes with zero steps) between two non-zero step events is less than a certain amount (say 30 minutes or an hour) then assume they were wearing the device in between. Can anyone suggest a more accurate estimate, or a reason why that estimate may be unreliable?
Thanks
