This is also relevant for airports

This gets my vote. This is one of the long standing annoyances I have with Arc/LocoKit’s processing engine.

This wasn’t a problem in Arc’s early years, because when you were inside buildings iPhones would produce very low accuracy location data, for example with reported accuracy of 60 to 100 metres. The way that the moving/stationary detection and visit edge detection algorithms work means that when location data accuracy is low it’s more likely to group those samples inside a single visit, because fine grained detail within the visit can’t be determined.

But what’s happened is that iPhones have steadily got better at determining accurate location inside buildings, and now report location data accuracy as good as 10 to 30 metres indoors in many cases. The data is cleaner, more accurate, and is reported as such.

That then feeds into Arc/LocoKit’s algorithms, allowing the algorithms to determine much more fine grained detail inside buildings, to the point where the algorithms no longer implicitly “know” that you’re inside a single building.

So, ironically, that better data makes Arc’s job harder for the cases where we do want all that stuff inside a large building to still be grouped together inside a single visit. It’s gone from being a problem that solved itself to being a problem that’s actually quite difficult to solve!

But I do have a bunch of ideas, and it is something I want to tackle eventually. So I’ll file this one in the “Planned” section, and hopefully it’ll get enough votes here that I can justify putting some serious development time into getting it working how we want.