Vp spatial accuracy

[tiffanychu90]

segment speeds

• Laurie added trip_instance_key for more robust joining between RT and schedule trips. See Slack and this release note. Update workflow to use trip_instance_key instead of bridge functions.
  • Remove functions no longer needed.
  • Don't make it backwards-compatible. Work through intermediate outputs to just add the new column or rename columns. Want Mar-Jul 2023 speeds to be similar.
  • Feb 2023 data hadn't fully worked through segmenting/time zone issues yet, and Jan 2023 data doesn't include time zone considerations, so it's not actually the right service_date.
• Re-download Feb 2023-Jun 2023 scheduled trips data because now there are significant changes, adding trip_instance_key, gtfs_dataset_key is present now. Other columns that have been fluctuating related to dealing with time zones and dates have stabilized and now we have a good set of columns we want to keep and use.
• Include scripts to patch in the additional columns now present in fct_vehicle_locations...like schedule_gtfs_dataset_key, _gtfs_dataset_name being renamed to gtfs_dataset_name.
• Housekeeping in GCS and move some of the intermediate outputs for normal/special into their own folders
• Epic - Daskify RT segment speeds #592

spatial accuracy

• use trip_instance_key to join vp with scheduled trips / shapes
• TODO: tweak this script because either dask.delayed is too slow or doing it on the dask cluster crashes it
• Research Request - Spatial accuracy metric #820