PostgreSQL query to detect overlapping time ranges

Question

I have a table in PostgreSQL 9.2 that looks like this (simplified):

CREATE TABLE my_features
(
  id integer NOT NULL,
  feature_id integer NOT NULL,
  begin_         


        

Answer 1

Here is an observation. If there are overlapping time periods for a feature, then at least one time period overlaps with the preceding one as defined by begin_time. (You can look at this the other way. If there are no such overlaps then there is always a gap between one time frame and the next and nothing overlaps.)

This leads to the following query for determining overlaps:

select f.feature_id
from (select f.feature_id,
             (case when lag(end_time) over (partition by feature_id order by begin_time) > begin_time
                   then 1 else 0
              end) as HasOverlap
      from my_features f
     ) f
group by f.feature_id
having max(HaxOverlap) = 1;