Better performance metrics

[radumas]

Issues of false positives aside, I think performance metrics representing the data we use these datasets for would be more appropriate than %links. e.g.:

• centreline: % total VKT
• HERE: % of weekday observations for a given month.

[chmnata]

For centreline 80.8% of 2016 AADT is matched up to SharedStreets:

SELECT 100.0* SUM(CASE WHEN ss.reference_id IS NULL THEN 0 ELSE aadt END) / SUM(aadt)
from uoft_volume.aadt_final 
LEFT join gis_shared_streets.centreline_matched ss ON centreline_id = geo_id::BIGINT and (case when direction = 'Northbound' or direction = 'Eastbound' then 1 else -1 end) = dir_bin
where yr=2016

70.75% when snapToIntersections=TRUE is used:

SELECT 100.0* SUM(CASE WHEN ss.reference_id IS NULL THEN 0 ELSE aadt END) / SUM(aadt)
from uoft_volume.aadt_final 
LEFT join gis_shared_streets.centreline_snap ss ON centreline_id = geo_id::BIGINT and (case when direction = 'Northbound' or direction = 'Eastbound' then 1 else -1 end) = dir_bin
where yr=2016

[chmnata]

For HERE 95.7% of weekday observations for 201903 when snaptointersection=TRUE is used:

SELECT 100.0* SUM(CASE WHEN ss.reference_id IS NULL THEN 0 ELSE 1 END) / count(1)
from here.ta_201903
LEFT join gis_shared_streets.here_matched_snap ss using (link_dir)
where date_part('isodow'::text, tx) < 6::double precision

For HERE 98.2% of weekday observation for 201903 when snaptoinetsection=FALSE is used:

SELECT 100.0* SUM(CASE WHEN ss.reference_id IS NULL THEN 0 ELSE 1 END) / count(1)
from here.ta_201903
LEFT join gis_shared_streets.here_matched_t6 ss using (link_dir)
where date_part('isodow'::text, tx) < 6::double precision