switch back to vp-sjoined-to-roads, aggregate shapes up to route-direction with stop_pairs

[tiffanychu90]

road segment speeds

• instead of cutting only portion of local roads each date, cut all the local roads at once and use for the entire year
• cut 1,000 m segments again, since interpolation at 250m showed challenges...from segment to segment, the vp it's selecting is jumping all over the place...not guarantee it's proceeding in order
• go back to sjoin method, and draw tighter buffers, and remove vp that aren't moving (vp_primary_direction=="Unknown"), and only allow vp that have moved since prior ping to join
• notebook to look at distribution of speeds where we can grab 2 points within a road segment...looks like stop segment speeds, so distribution is ok.
  • we do have a lot of segments where we do have just 1 vp. but, if the prior vp joined segment several back, how do we know bus hasn't traveled along other roads and them come back to this road?
• Epic - Transit speeds on road segments #916, Research Request - Improve sjoin of vp to segments by including cardinal direction  #921

gtfs_funnel / aggregation of shapes to route-direction

• way to deal with variation between shapes when aggregating to route-direction is to add a stop_pair column, which tells us the pair of prior_stop_id and current_stop_id, instead of using stop_sequence. when we aggregate, we only want to aggregate segments that are traveling between the same 2 stops
  • add new column in stop_times_with_direction.py, switch from dask to pandas because it's faster
• in multiday_avg.py, add a second aggregation where a week's worth of speeds by stop segments can be aggregated to route-direction-segments.
• in 26_weekly_avg_speeds.ipynb take a look at example shape and make bar chart of speeds
  • TODO: continue exploratory work later, maybe use portfolio to get eyes on what's happening for each operator
• Research Request - Roll-up single day speeds to weekday vs weekend + peak vs offpeak average speeds  #960

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/26_weekly_avg_speeds.ipynb
• rt_segment_speeds/scripts/speeds-from-road-sjoin.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/26_weekly_avg_speeds.ipynb
• rt_segment_speeds/scripts/speeds-from-road-sjoin.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/26_weekly_avg_speeds.ipynb
• rt_segment_speeds/scripts/speeds-from-road-sjoin.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/26_weekly_avg_speeds.ipynb
• rt_segment_speeds/scripts/speeds-from-road-sjoin.ipynb

[github-actions]

nbviewer URLs for impacted notebooks:

• rt_segment_speeds/26_weekly_avg_speeds.ipynb
• rt_segment_speeds/scripts/speeds-from-road-sjoin.ipynb