Research Request - GTFS Digest Metrics Shakedown

[tiffanychu90]

Complete the below when receiving a research request, and continue to add to this issue as you receive additional details and produce deliverables. Be sure to also add the appropriate project-level label to this issue (eg gtfs-rt, DLA).

Research Question

Single sentence description: Draft GTFS digest is up, now decide which metrics are worth charts (time-series), which are more easily captured by tables (perhaps mostly static with little changes)

• Epic - GTFS Digest (exploratory portfolio) #989

Tables

Grain

• route-direction
• in digest, do not use schedule_gtfs_dataset_key or name (schedule), but use organization_source_record_id and organization_name.
• organize by district
• meaningful route name displayed, and charts use this standardized route_name, not route_id:
  • put through script to standardize / parse route_id over time so we can identify the same route over time (route_id2).
  • further clean up by operator so that combo route_short_name and route_long_name does not have redundancy (1 Route 1 -> Route 1)

Operator Stats ("digest/operator_profiles" and "digest/operator_routes")

• monthly scheduled service hours by day of week (day_type, time_of_day)
• number of routes, service area (length in miles, only count each route-direction once)
• number of stops served, total stop arrivals, arrivals per stop
• route typology breakdown

Schedule ("digest/operator_schedule_rt_category"?)

• avg scheduled service minutes
• avg stop distance (meters) -> change to per mile?
• all_day / peak / offpeak scheduled n_trips
• all_day /peak / offpeak scheduled frequency
• monthly scheduled service hours by day of week (day_type) and time_of_day

Speeds ("digest/segment_speeds")

• summary speeds for all_day /peak / offpeak

RT vs Schedule

• trips breakdown by sched_vp_category (10 trips schedule only, 100 trips schedule_and_vp, 5 trips vp only)
  • .drop_duplicates() but keep some dates where this distribution changes
• route breakdown by sched_vp_category ("digest/schedule_vp_metrics") don't do this, because routes can have different numbers of n_scheduled_trips, n_vp_trips, and we can assign incorrectly.

Charts

Schedule

• monthly scheduled service hours by day of week (day_type) and time_of_day
• number of trips by sched_vp_category?

Speeds

• summary speeds for all_day / peak / offpeak

RT vs Schedule

• filter to schedule_vp_category = "schedule_and_vp"

• vp per minute (goal line = 2)
• % vp in scheduled shape (goal line = 100%) - use all_day
• % RT journey with 1+/2+ vp (goal line = 100%) - use all_day, one chart shared for 1+ and 2+
• % schedule journey with 1+/2+ vp (goal line = 100%) - use all_day, one chart shared for 1+ and 2+
• comparison of scheduled to RT trip time (aggregated to route-direction)
  • breakdown n_trips early (5+ min early) / on-time within 5 min early or late / late (5+ min late)

Maps

Deliverables

• Parameterized notebook with supporting scripts (charting, report styling, or table aggregations).
  • Plug and play approach. How the metrics are displayed and organized potentially can get reshuffled, so set up functions for making each chart and then wrap them up for display.
• If we want to have dropdown by route, would altair's extract data help with displaying tables? Right now, tables are not filtered interactively, but that is desired.
• Try out great_tables, potentially can help with displaying tables. Their nanoplots look cool, but only take polars dfs, not pandas and df might have to be wide?

Data Sources

• Here