Перечень тегов:
#sql - SQL
#window - оконные функции

#sql #window
Вот такой вопрос появился на канале @data_engineerette

Моё решение такое:

SELECT event_type, date_start, date_end
FROM (
  SELECT event_type, 
  FIRST_VALUE(date) OVER (PARTITION BY event_mark ORDER BY date) AS date_start,
  LAST_VALUE(next_date) OVER (PARTITION BY event_mark ORDER BY date) AS date_end,
  last_event
  FROM (
      SELECT event_type, date, next_date, last_event, 
      sum(cng_event) over (order by date rows between unbounded preceding and current row) AS event_mark
      FROM (
        SELECT event_type, date,
        LEAD(date, 1, NULL) OVER (PARTITION BY 1 ORDER BY date) AS next_date,
        CASE WHEN LAG(event_type, 1, 0) OVER (PARTITION BY 1 ORDER BY date) = event_type THEN 0 ELSE 1 END AS cng_event,
        CASE WHEN LEAD(event_type, 1, 0) OVER (PARTITION BY 1 ORDER BY date) <> event_type THEN 1 ELSE 0 END AS last_event
        FROM events
        ORDER BY date) AS inn
    ) AS outt) AS final_tab
WHERE last_event = 1
ORDER BY 2;

Поиграться можно здесь или здесь