Hello,
Seems such started earlier this year or somewhat recently, perhaps due to some recent version or some changes to iOS.
Maybe it started in Feb. Or that’s where I started having some issues. I have a few days where there are a thousand plus brief stops all along some driven path, maybe some also at home if I recall. Been using Arc Recorder as well for some time. Not sure what it is or if there is some easy way to fix such days?
Hmm. That’s a weird one.
Is it in built up city areas, where buildings might be blocking line of sight to GPS satellites? Or is it areas where there should be clean and high accuracy GPS signal?
Oh and does the processing engine manage to clean it up on its own? Or do you need to do some manual edits to get it cleaned up?
I forgot how many I had, perhaps just a few and now looking that I can tell just one remaining that I before I didn’t feel like trying to fix at the time. Seems indeed it fixes itself but just takes a long time.
Area is suburbs with little but houses, commercial stores, maybe no buildings more than a few floors. Yet it is the whole trip such as this example of being on a freeway, with samples varying between 1 and a few seconds:
Maybe I was using some beta OS at the time. Since it’s been some months I’m not sure why it happened and it was only a handful of times now that I think about it. Just wanted to mention and not sure if it was from a recent version or iOS or something else. Perhaps I was just concerned that it might happen again but it hasn’t.
Of other abnormalities, there is also one case (it seems) where I am at home yet it shows something like (since I don’t have it in front of me) travel of some feet away, as if it is just getting different coordinates yet the whole time I’m stationary and I haven’t yet tried to fix such.
Just some weird things that are rare. Related to such, from what I recall often I can confirm all yet I can’t with home which I it seems I sometimes need to confirm manually.
Apologies for grouping stuff together. Another issue from what I recall, perhaps known, sometimes at a location that might be a park or a big store, there is variation between days of either some visit there, walking inbetween, another visit, or just one location of some duration. Perhaps I’d prefer it showing being in one location such as a big store, park, or mall for such a duration. I haven’t tried recently but from what I recall it is not easy to delete lets say walking in between to just show it as one visit.
Wow, that screenshot is CRAZY! I’ve never seen anything even remotely like that! That’s… just wow. It’s completely rhythmic, like a metronome 
That must surely have been some iOS glitch. The fact that it’s such a perfect pattern means it can’t possibly have been artefacts of real world data.
Yeah that would make sense. Though I’m always on the betas, and I’ve never seen ANYTHING like it ever. So that’s quite a rare gem of a glitch you’ve got there 
It is possible for something similar (though far more naturally random looking) happening in built up city areas, where location data accuracy can range from 30 to 100 metres or worse. When travelling slowly (eg walking, or car stuck in traffic) the poor location data accuracy makes the stationary/moving state detection next to impossible. So it can frequently fall into thinking “well I guess they must be stationary; shrug” because there’s not enough detail of data to recognise confident movement.
This one is unfortunately quite common. The exact cause could be a number of things, but it’s the same basic situation again: poor location data accuracy, that fluctuates over time.
The way it typically manifests is as drifting location while you’re definitely stationary inside a building. like the path drifting off in the same direction, potentially even hundreds of metres, then drifting back again. Then it all repeats again some minutes or hours later.
It could be due to the phone not having line of sight to enough GPS satellites, and the ones that it can see are moving away, out of sight, causing a weird gradual offsetting over a few minutes. That’s one of my hunches anyway.
Arc has a built in system for learning to detect that and compensate for it, called “Trust Factor”. Whenever you correct drifting data inside a visit, marking it stationary, the Trust Factor system learns “ok so at this location, real stationary data might present with actual movement, at certain speeds”. Then it knows to lower its trust in the incoming data at that location.
Like, the phone might be reporting “here’s location data with 30 metres accuracy”, but Trust Factor has learnt that it’s likely a lie, and responds “ok, thanks buddy, but I’m gonna mark that as 130 metres accuracy”. Then the filtering algorithms know to treat it as more suspicious, and adapt to that.
So basically the solution there is: if you get drifting data inside a building, mark it as stationary, then Trust Factor will learn to pass on the message to the filtering systems in future. End result is… less drifting data showing up in the visits!
No worries about grouping!
For that issue, yeah that’s a known weakness in Arc’s timeline processing. I’ve been mulling over ways to improve it for literally years. Hopefully soon I’ll get a chance to work on them, now that app development is AI Powered 
. It’s much easier to get more work done these days, so I have more hope for many of the things that were previously filed under “that would be great, but I’ll never get time”. Now it’s more “Claude and I could churn that out in an afternoon”. So fingers crossed I get to that one soon! It’s on many people’s wish lists.
Yep. Sometimes you can clean it up the way you want, but often you simply can’t. Which is why it really does need a solution inside Arc’s processing engine. Arc needs to be smarter in those cases.
Thank you very much for the detailed response and great help as always.
