search

transitmatters / mbta-performance

For processing performance data for the data dashboard
MIT License
1 stars 1 forks source link

headways data quality? #10

Open hamima-halim opened 7 months ago

hamima-halim commented 7 months ago

the scheduled headways in LAMP are pretty irregular. in gobble, we disregard the reported scheduled headways entirely and recalculate our own from gtfs.

THAT BEING SAID.... the headways we calculate seem to be identical the ones we report in gobble, and if we recalculate the lamp events in the same was we get the same numbers. but also, i might be illiterate.

devinmatte commented 7 months ago

We also process the monthly data files: https://github.com/transitmatters/t-performance-dash/blob/main/server/rapid/process_events.py

There may be some logic in here worth looking at (looks like we don't touch gtfs in this, so maybe not)

devinmatte commented 7 months ago

Aha I knew there was something: https://github.com/transitmatters/t-performance-dash/blob/main/server/bus/bus2train.py

We do this for bus monthly data we ingest (but for some reason not for train data, may want to update that)

devinmatte commented 6 months ago

We should look into smoothing headways into 15 or 30 minute averages to avoid these jumps in the bar