Hi!

Arc already does this automatically. Every time you confirm or correct the activity type of a trip, Arc uses that information to improve its knowledge of your preferred modes of transport, common routes, common times of day, and much more. So the activity types list you see in the confirm/edit view is sorted specifically for you, based on your previous confirmations and corrections, and based on that specific trip.

So that means that if you always take the bus to work, along the same route, Arc will auto detect bus for that route each day. And if for some reason it doesn’t pick bus as the first choice on one day, it will still probably have bus next in the list, as the next most likely choice.

Activity types that you never use, or that are a very bad match for the specifics of the trip, will always be near the bottom of the list.

So in a sense this feature request is “already implemented”, because Arc does it automatically. As long as you confirm and correct the activity types for trips, when Arc asks for confirmation (or when Arc got it wrong), the types list will be automatically sorted to match you personally.

Oh gotcha! Maybe it’s confused because I’m a work-from-anywhere freelancer so while I do tend to stick to the same bus lines, I don’t take the at the same time.

Also, I deleted the app for a few weeks (and then it got that awesome ginormous update!), so perhaps that is influencing odd choices of locations and transit modes? E.g. back when we couldn’t add a new private place from a specific activity within a longer segment, I made a couple of “catch all” locations. Thus they cover a lot of physical area, but they haven’t been used since last year. Yet upon reinstall, Arc has been assigning those wide-area locations instead of “home”, which would be the most frequent location by far. Mayhaps I just need to let the restore finish up!

Yep, while the restore is happening Arc can get much more easily confused, because it doesn’t have all of the data back yet that it had previously learnt from. So as the restore progresses, it will gradually get back its old knowledge (rebuilding its ML models as restored data arrives). It might need a bit more babying during the restore, while it makes dumber than usual decisions.

Also yeah, if you’re a work-from-anywhere kind of person, then Arc will have a harder time with activity type predictions. I’m the same, in that I tend to work different hours each day, and from various cafes depending on my mood, and each day doesn’t stick to any strict daily routine. And some days I’ll just randomly head off to a different part of town, to do some impulse exploring. So sometimes Arc knows exactly what I’m doing, and sometimes it gets caught out with me doing something that doesn’t fit any previous patterns, or could easily match up with multiple different common patterns, with no single obvious best match.

One example is when I go to my local train station. To get there I might walk, or take a shuttle bus, or ride an e-scooter, or by bicycle, by tuktuk, by motorbike taxi, or by songthaew (my condo’s shared taxi). So sometimes Arc can easily spot which activity type it is for that trip, but sometimes it’s a really tough call. Although it will still always put all of those activity types at the top of the list, even if it didn’t manage to pick exactly the right one as the best match.

Ha! Yeah, a couple of days ago it decided that instead of spending the night at home, I was at a nearby fast food joint for over 12 hours! What an unhealthy lifestyle I have.

This is all super interesting info on how this works! I’ll let it restore and enjoy the dumbness in the meantime. :D