Speed graph very inaccurate

Hi Brenton,

We have been looking at this and There are a couple of problems here, all on our side, due to the inaccuracies of GPS tracking. The differences in what you see reported across applications are from what each of them does in processing the raw data: specifically, removing outliers and anomalies. We don't yet have a good algorithm to do this on ZED9, though we clearly need one!

Thanks for your patience. I do want to fix this as soon as we can figure out a good way.

Olivia

We just launched an update with some improvements in how we're smoothing the data to calculate max speed. While not an exact match with the 21.5mph you saw in MotionX, your workout 466 on ZED9 now shows a max speed of 22.1mph.

We don't yet have the capabilities to address the remaining problems around the GPS data highlighted by workout 384. The new max speed algorithm did cut down the reported max on workout 384 to 45.5mph, but there is still a spike in the speed graph where the app recorded a big jump in your location. Though smoother, we are also still reporting some of the recorded movement while you were staying put at lunch. We hope to continue to continue improving our handling of GPS anomalies over time. Hopefully today's changes help for the interim.

Thanks for hanging in there with us Brenton! The speed and duration issues you've been hitting are so far unique to the iPhone tracking apps.

The speed spike at the beginning of your ride 384 may not be an EveryTrail issue though. What most likely happened is that the GPS accuracy was low when you first started the app so it placed you in a location relatively far from where you actually were. When it locked onto more satellite signals for higher accuracy, your location changed significantly, fast. Cleaning the data to remove anomalies helps when there are just a few points out of line, but there's not much we can do with real data that just happens to be wrong.

I've been looking further into the idling issue you saw. While it is particular to the way EveryTrail handles idling, I'm not actually convinced it's a bug, or at least not something we should auto-correct: does staying in one place always mean you're idle? What if you break to do a set of pushups? You may still want to consider this part of your workout time. Some software (e.g., Garmin's) auto-pause when you stop moving, but that makes it hard to track mixed workouts (which is also how we originally accumulated a big pile o' tracking devices).

If you go back to EveryTrail, two options that may help are 1) pause while stopped (they actually call it Stop, but you can resume) or 2) you could try increasing the distance filter in the app's GPS settings so that it doesn't record a new location unless you travel further than the default 10m.

Thanks for the great feedback and ideas, Brenton!

It's interesting and good to know that the app was paused. I looked through the raw GPS file and didn't see any significant time breaks -- it actually looks like the app kept recording your location even while paused! That would explain why that time shows up on ZED9. So far you're the first to encounter this, but I'll take another look and see if I can find a marker we can use to exclude that data.

We do intend to continue improving our overall data anomaly handling over time, but in the short term we don't have a good, scaleable solution that we can implement.

Though it's by no means a great experience, there is one workaround available now: you can always manually edit the source file on your computer. I actually did this last night when I found that I had forgotten to stop my Garmin after my run in the morning. It only took a minute, but some comfort with looking at structured data files definitely helps.

Thanks again for your continued support and feedback. I love seeing your rides on ZED9!

Olivia

I've found the problem now. Thanks Brenton! The root issue is actually non-trivial so it may take us a bit of time, but I will keep you posted.

Hi Brenton,

We applied a new duration algorithm for .gpx files today to ignore inactive time. We reprocessed all of your existing workouts and, though not an exact match with your other tracking programs (see below), the numbers are now much closer to what you were expecting for active time.

Durations for the workouts you listed the MotionX stats for above are registering as a little bit shorter on ZED9. This surprised me but in looking closer at the raw data I do believe this is correct. Using workout 496 as an example:
Your trip out was from 21:05.43 to 21:11.06, or 5:23 total.
Your return trip was from 21:46.40 to 21:51.16, or 4:36 total.
We removed another 35 seconds of time in which you were stationary (stoplights?) from that 9:59 travel time, for a new duration on ZED9 of 9:24. I am still not sure what MotionX was counting to get to 11:43.9, but at least they are now in the same ballpark.

Please let me know if you have any more questions!

Thanks,
Olivia

We do still have a bug around workout pauses in some situations. Unfortunately EveryTrail seems to trigger it more often -- not sure why. We will keep trying to track this down and keep you posted. Thanks for your patience.