SummerProject2026/sql/local/21_build_episodes.sql
2026-07-15 17:55:53 -06:00

116 lines
3.7 KiB
SQL

-- Construct episodes across the complete, globally ordered input history.
-- A gap exactly equal to the configured threshold remains in the same episode.
CREATE OR REPLACE TABLE sequenced_events AS
WITH with_previous AS (
SELECT
*,
lag(event_ts) OVER (
PARTITION BY vehicle_token
ORDER BY event_ts, internal_event_id
) AS previous_event_ts
FROM clean_events
),
with_markers AS (
SELECT
*,
CASE
WHEN previous_event_ts IS NULL THEN 1
WHEN event_ts - previous_event_ts
> (SELECT episode_gap_days FROM build_config)
* INTERVAL '1 day'
THEN 1
ELSE 0
END AS starts_episode
FROM with_previous
),
with_episode_number AS (
SELECT
*,
sum(starts_episode) OVER (
PARTITION BY vehicle_token
ORDER BY event_ts, internal_event_id
ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW
) AS episode_number
FROM with_markers
)
SELECT
*,
row_number() OVER (
PARTITION BY vehicle_token, episode_number
ORDER BY event_ts, internal_event_id
) AS attempt_number,
row_number() OVER (
PARTITION BY vehicle_token, episode_number
ORDER BY event_ts DESC, internal_event_id DESC
) AS reverse_attempt_number
FROM with_episode_number;
CREATE OR REPLACE TABLE episode_daily_activity AS
SELECT
vehicle_token,
episode_number,
cast(event_ts AS DATE) AS event_date,
count(*) AS events_in_day
FROM sequenced_events
GROUP BY vehicle_token, episode_number, cast(event_ts AS DATE);
CREATE OR REPLACE TABLE episode_aggregates AS
SELECT
e.vehicle_token,
e.vehicle_bucket,
e.episode_number,
min(e.event_ts) AS episode_start,
max(e.event_ts) AS episode_end,
count(*) AS attempt_count,
count(e.canonical_outcome) AS labeled_attempt_count,
coalesce(bool_or(e.canonical_outcome = 'pass'), false) AS eventually_passed,
max(d.events_in_day) AS max_events_in_day,
(list(e.observed_make ORDER BY e.event_ts DESC, e.internal_event_id DESC)
FILTER (WHERE e.observed_make IS NOT NULL))[1]
AS episode_last_observed_make,
(list(e.observed_model ORDER BY e.event_ts DESC, e.internal_event_id DESC)
FILTER (WHERE e.observed_model IS NOT NULL))[1]
AS episode_last_observed_model,
(list(e.observed_model_year ORDER BY e.event_ts DESC, e.internal_event_id DESC)
FILTER (WHERE e.observed_model_year IS NOT NULL))[1]
AS episode_last_observed_model_year
FROM sequenced_events AS e
JOIN episode_daily_activity AS d
ON d.vehicle_token = e.vehicle_token
AND d.episode_number = e.episode_number
AND d.event_date = cast(e.event_ts AS DATE)
GROUP BY e.vehicle_token, e.vehicle_bucket, e.episode_number;
CREATE OR REPLACE TABLE inspection_episodes AS
SELECT
a.vehicle_token,
a.vehicle_bucket,
a.episode_number,
a.episode_start,
a.episode_end,
f.internal_event_id AS first_internal_event_id,
f.source_era,
f.public_county,
f.canonical_outcome AS first_outcome,
f.outcome_label_source AS first_outcome_label_source,
f.program_type AS first_program_type,
f.test_type AS first_test_type,
l.canonical_outcome AS final_outcome,
a.attempt_count,
a.labeled_attempt_count,
a.eventually_passed,
a.max_events_in_day,
a.episode_last_observed_make,
a.episode_last_observed_model,
a.episode_last_observed_model_year
FROM episode_aggregates AS a
JOIN sequenced_events AS f
ON f.vehicle_token = a.vehicle_token
AND f.episode_number = a.episode_number
AND f.attempt_number = 1
JOIN sequenced_events AS l
ON l.vehicle_token = a.vehicle_token
AND l.episode_number = a.episode_number
AND l.reverse_attempt_number = 1;