Data
One month, four files.
A real agency's September: vehicle location, passenger counts, fare taps and the published timetable. Summarised here, never shared raw.
AVL
Vehicle location
Where every bus is, and how late. 5.5 million stop events.
APC
Passenger counting
Who got on and off at each stop. 5.2 million stop events.
AFC
Fare collection
Every tap and ticket, including the cross-agency transfers. 1.7 million taps.
GTFS
Timetable
The published schedule: the one open, public input.
Cross-checks
Do the keys line up across the four files? Mostly not.
Other systems
Maintenance, charging and whatever else a party runs.
Four files from one agency, and the keys already disagree. The trip identifier in operations is not the trip identifier in the timetable. That is why the exchange owns semantics.