I recently used this technique as a quick and dirty way to find
anomalies in time series data and thought I would share. Let’s say you
have simple time series data, event counts grouped by day: SELECT
TO_CHAR(DATE_TRUNC('week', updates.date),'YYYY-MM...