Quote:
Originally Posted by TransitZilla
I can assure you that there is a team that is working hard on these issues. The current arrival prediction system has a few limitations, including:
-it is unable to provide real-time predictions when a bus is "off-route" (e.g. on a detour).
-it relies on the scheduled runtime of routes to predict arrival times and cannot account for live traffic conditions. (e.g., if the schedule says the bus will take 5 minutes to get from its current location to your stop, the prediction will be 5 minutes).
-accuracy is lower for trips that have not yet started because even if we know the location of the bus, we cannot know for certain when it will depart from the terminus.
The team is aware of an issue that affect trips that start at Parliament station, such as those on route 19. This issue stems from a data problem that arises when schedules are loaded from the scheduling system to the dispatch system. These systems are supplied by 2 different vendors. Work is underway to determine a solution.
The result of the problem is that as buses on route 19 are deadheading between trips (to turn around to head back east) the buses are considered "off-route" and the prediction system stops providing real-time predictions for them. This may be causing the issues you see in the Next Departures tool on the website and on the screens.
The best workaround to this is to use Transit or another app that will show you the GPS location of the buses on a map. There is not very much interlining on route 19; nearly every trip that arrives as an inbound trip on route 19 turns around and departs as an eastbound route 19 a few minutes later. Transit (and other apps) will show you the GPS location of all buses on the route. In Transit, you can check the map for both directions on route 19; if you see a bus on Kent St under either direction, it means it is laying up and will start its next eastbound trip soon.
I do know that some issues with the Next Departures tool (such as when it incorrectly shows that there are no departures for the next 4 hours) have been challenging for staff to reproduce. Feel free to PM me with examples if you see this happen. Details like the exact time you refreshed the data, device OS, browser, etc would be helpful to uncover the cause of that issue.
|
These issues aren't getting any better.
If anything, they're getting worse.
Another example from last night: The 19 didn't have a first stop problem. It had first TWO stop problem. It didn't show up at stop 3052 as scheduled, which is to be expected these days because of the Porter Island detour. No huge deal; it's what riders on the 19 have, sadly, had to come to expect.
And it also wasn't visible on Transee at the usual layup spots. No big deal, because Transsee did show me a 19 Parliament making its way past U of O, and I guessed, correctly, that it would turn around and start its eastbound trip, late relative to "the schedule", which no longer exists, to be sure, but eventually.
HOWEVER: at no time did OC Transpo's own web site give me a passing time for that trip at Parliament 3052, or at Metcalfe 3008. It DID, however, give me an accurate time for Mackenzie King.
It was still not recognizing that there was a bus in service an on its way at 3052 or 3008 even after the bus was in service, in motion, and visibly on its way towards the stop. OC Transpo's own web site could only tell me the next scheduled trip, not the actual trip that was already in service. But it was only "blind" to that trip at 3052 and 3008, and not at stops further down the line.
Nor was the bus visible to Transsee any more, even though it had been visible before it turned around.
Also: For the past two or three weeks, OC Transpo's website has been overly pessimistic in projecting the next passing times of buses already in motion, projecting them to arrive two or three minutes after they actually do... which misleads riders into heading to stops or attempting to make connections with buses that have already come and gone.
The data service is an absolute goddamn mess, and is no better than a Magic 8 Ball at this point. As was the case last night, you are better off ignoring what OC Transpo's data tells you much of the time, which really should not be the case, to put it mildly.
It's one thing for "the team" to be collecting examples or real-world reports from users, but CRAZY IDEA ALERT - how about "the team" actually go out in to the world and ground-truth the system themselves? Because I am not convinced at all that "the team" truly understands the problems, the extent of the problems, and the impact on riders.
I would also love to know where the Transit Commission is on all this stuff, because I don't see much evidence that they are on top of it at all.