Group chronologically - sql

Question

I'm trying to get the top result from each status from a table grouped by customer id and status, ordered by time.
The data given:
CustNo Date Status
1 2016-03-24 C
1 2016-02-08 C
1 2016-01-17 A
1 2015-12-04 C
2 2016-04-28 B
2 2016-03-25 C
2 2016-02-13 C
2 2016-01-04 C
3 2016-02-02 A
3 2016-01-09 A
3 2015-12-12 A
3 2015-11-30 A
I want the output to look like this:
CustNo Date Status
1 2016-03-24 C
1 2016-01-17 A
1 2015-12-04 C
2 2016-04-28 B
2 2016-03-25 C
3 2016-02-02 A
As you can see I want the top date for each status change (if any) within each customer. I solved it for customer 2 and 3 where there is no change of status or the status never changes back, but as for customer 1 the status has changed from C to A and back to C and this is the tricky part (for me at least). I always seem to get the C status grouped all together.

Answer 1

You can try this.
With CTE as
{
select CustNo,
DATE,
Status,
ROW_NUMBER () over (partition by CustNo order by date) as ord
from tab
)
SELECT CustNo,
DATE,
Status
FROM CTE
EXCEPT
SELECT C1.CustNo,
C1.DATE,
C1.Status
FROM CTE C1
INNER JOIN CTE C2 ON C1.CustNo = C2.CustNo
AND C1.Status = C2.Status
AND C1.ord + 1 = C2.ord

Answer 2

You could use windowed function ROW_NUMBER to calculate groups:
WITH cte AS (
SELECT *,
ROW_NUMBER() OVER(PARTITION BY CustNo ORDER BY Date) -
ROW_NUMBER() OVER(PARTITION BY CustNo, Status ORDER BY Date) AS grp
FROM mytable
)
SELECT CustNo, Status, MAX(Date) AS Date
FROM cte
GROUP BY CustNo, Status, grp
ORDER BY CustNo, Date DESC;
LiveDemo
Output:
╔════════╦════════╦═════════════════════╗
║ CustNo ║ Status ║ Date ║
╠════════╬════════╬═════════════════════╣
║ 1 ║ C ║ 24.03.2016 00:00:00 ║
║ 1 ║ A ║ 17.01.2016 00:00:00 ║
║ 1 ║ C ║ 04.12.2015 00:00:00 ║
║ 2 ║ B ║ 28.04.2016 00:00:00 ║
║ 2 ║ C ║ 25.03.2016 00:00:00 ║
║ 3 ║ A ║ 02.02.2016 00:00:00 ║
╚════════╩════════╩═════════════════════╝