SQL - Group By based on sequence - Oracle/Postgre - sql

Question

I need some help.
To perform a group by based on a sequence.
I'm using Oracle or Postgres.
I have the following scenario. The ID_SEQ is based on equip_id, Day and Stat. Creating a sequence.
I need to group the intervals between these sequences.
Example:
EQUIP_ID DAY STAT DATE ID_SEQ
JSTD123 19/06/2017 ON 19/06/2017 16:39 1
JSTD123 19/06/2017 OFF 19/06/2017 16:41 1
JSTD123 01/07/2017 ON 01/07/2017 13:50 1
JSTD123 01/07/2017 OFF 01/07/2017 13:51 1
JSTD123 01/07/2017 OFF 01/07/2017 14:40 2
JSTD123 01/07/2017 ON 01/07/2017 15:20 1
JSTD123 01/07/2017 ON 01/07/2017 15:20 2
JSTD123 01/07/2017 ON 01/07/2017 15:22 3
JSTD123 01/07/2017 ON 01/07/2017 15:22 4
JSTD123 01/07/2017 ON 01/07/2017 15:23 5
JSTD123 01/07/2017 ON 01/07/2017 15:26 6
JSTD123 01/07/2017 ON 01/07/2017 15:26 7
I would like to have the following result:
EQUIP_ID DATE STAT START END
JSTD123 19/06/2017 ON 19/06/2017 16:39 19/06/2017 16:39
JSTD123 19/06/2017 OFF 19/06/2017 16:41 19/06/2017 16:41
JSTD123 01/07/2017 ON 01/07/2017 13:50 01/07/2017 13:50
JSTD123 01/07/2017 OFF 01/07/2017 13:51 01/07/2017 14:40
JSTD123 01/07/2017 ON 01/07/2017 15:20 01/07/2017 15:26
I can't get a similar output.

Answer 1

I think this is what you are trying to do. Grouping consecutive rows with the same stat on a given day and getting the min date and max date of that group.
The logic is to assign groups by getting the previous value of stat (per equip_id and day) using lag and then using a running sum to reset when a new stat value is encountered. After this group assignment is done, you can just use group by to get the min and max date per equip_id,stat,day and grp.
SELECT equip_id,
day,
stat,
min(date),
max(date)
FROM
(SELECT t.*,
sum(col) over(partition BY equip_id,day ORDER BY date) AS grp
FROM
(SELECT t.*,
CASE WHEN stat=lag(stat) over(partition BY equip_id,day ORDER BY date) THEN 0 ELSE 1 END AS col
FROM t
) t
) t
GROUP BY equip_id,day,stat,grp
Sample Demo