tripy and blocky

Work this past year has concentrated on identifying revenue vs. non-rev (dead-head) trips and trip segments. The problem for TriMet is that information not intended for public consumption (e.g., a stop that is used to trigger an overhead sign change) shows up in our GTFS. If this information is not filtered, and thus is shown on say a stop page http://trimet.org/ride/stop.html?id=11939, a customer might be led to believe that a bus line is assigned to this stop. It gets bad when said customer uses this information and visits the stop, dutifully waiting for their bus, which eventually streams by on it's way back to the garage (e.g., in a trip or two, the bus stop 11939 was simply a trigger point for an overhead sign change for line 62 ... the bus never actually stopped at 11939, nor even pulled over to the stop).

Anyway, I've added code that uses block and trip information to identify trips and stop times that are dead-heading and out of revenue. With these routes/stops/times identified, bad information is then filtered out of route stop lists and stop time schedules.