Reputation: 2750
I have these data:
year code points
-------------
2017 M 1
2018 L 3
2019 L 5
I need the total of points of 2019, taking all codes before 2019 that match the code of 2019
so the result should be:
y c p
----------
2019 L 8
I have an oracle database, not sure how to do it (with 'connect by' ?)
To test the resquest easily:
WITH test AS (
SELECT 2017 y, 'M' code, 1 point FROM DUAL
UNION
SELECT 2018 y, 'L' code, 3 point FROM DUAL
UNION
SELECT 2019 y, 'L' code, 5 point FROM DUAL
)
SELECT * FROM test
Upvotes: 0
Views: 33
Reputation: 35900
You can do a self join
to achieve the desired result as follows:
SQL> WITH test AS (
2 SELECT 2017 y, 'M' code, 1 point FROM DUAL
3 UNION
4 SELECT 2018 y, 'L' code, 3 point FROM DUAL
5 UNION
6 SELECT 2019 y, 'L' code, 5 point FROM DUAL
7 )
8 SELECT
9 T1.Y,
10 T1.POINT + COALESCE(T2.POINT,0) AS POINTS
11 FROM
12 TEST T1
13 LEFT JOIN TEST T2 ON T1.CODE = T2.CODE
14 AND T1.Y > T2.Y
15 WHERE T1.Y = '2019'
16 ;
Y POINTS
---------- ----------
2019 8
SQL>
Or you can use the analytical function
as follows:
SQL> WITH test AS (
2 SELECT 2017 y, 'M' code, 1 point FROM DUAL
3 UNION
4 SELECT 2018 y, 'L' code, 3 point FROM DUAL
5 UNION
6 SELECT 2019 y, 'L' code, 5 point FROM DUAL
7 )
8 SELECT Y, POINTS FROM
9 (SELECT
10 T1.Y,
11 SUM(T1.POINT) OVER (PARTITION BY CODE ORDER BY Y) AS POINTS
12 FROM
13 TEST T1)
14 WHERE Y = '2019'
15 ;
Y POINTS
---------- ----------
2019 8
SQL>
Upvotes: 1
Reputation: 1269773
I think I would go for:
select sum(points)
from (select code, sum(points) as points
from t
where y <= 2019
group by code
having max(y) = 2019
) x;
This handles two edge cases:
Upvotes: 0