寻找性能有效的SQL代码,根据DATE Sum计算Rolling Sum。我的数据如下:
+----+-----------+----+
| ID | Date | Qt |
+----+-----------+----+
| 1 | 1/12/2009 | 2 |
| 2 | 1/13/2009 | 3 |
| 3 | 1/14/2009 | 10 |
| 4 | 1/15/2009 | 2 |
| 5 | 1/16/2009 | 3 |
| 6 | 1/17/2009 | 7 |
| 7 | 1/18/2009 | 5 |
| 8 | 1/19/2009 | 4 |
| 9 | 1/20/2009 | 2 |
| 10 | 1/21/2009 | 3 |
| 11 | 1/22/2009 | 10 |
| 12 | 1/23/2009 | 2 |
| 13 | 1/24/2009 | 3 |
| 14 | 1/25/2009 | 7 |
| 15 | 1/26/2009 | 5 |
| 16 | 1/27/2009 | 4 |
| 17 | 1/28/2009 | 1 |
| 18 | 1/29/2009 | 0 |
| 19 | 1/30/2009 | 8 |
| 20 | 1/31/2009 | 9 |
+----+-----------+----+
输出应该是:
+----+-----------+----+-----------+
| ID | Date | Qt | Roll10Day |
+----+-----------+----+-----------+
| 1 | 1/12/2009 | 2 | 2 |
| 2 | 1/13/2009 | 3 | 5 |
| 3 | 1/14/2009 | 10 | 15 |
| 4 | 1/15/2009 | 2 | 17 |
| 5 | 1/16/2009 | 3 | 20 |
| 6 | 1/17/2009 | 7 | 27 |
| 7 | 1/18/2009 | 5 | 32 |
| 8 | 1/19/2009 | 4 | 36 |
| 9 | 1/20/2009 | 2 | 38 |
| 10 | 1/21/2009 | 3 | 41 |
| 11 | 1/22/2009 | 10 | 49 |
| 12 | 1/23/2009 | 2 | 48 |
| 13 | 1/24/2009 | 3 | 41 |
| 14 | 1/25/2009 | 7 | 46 |
| 15 | 1/26/2009 | 5 | 48 |
| 16 | 1/27/2009 | 4 | 45 |
| 17 | 1/28/2009 | 1 | 41 |
| 18 | 1/29/2009 | 0 | 37 |
| 19 | 1/30/2009 | 8 | 43 |
| 20 | 1/31/2009 | 9 | 49 |
+----+-----------+----+-----------+
我们可以使用窗口功能实现吗?我知道我可以在SQL中使用内部查询。
答案 0 :(得分:1)
通常使用带有sum()
的窗口子句来处理:
select t.*,
sum(qt) over (order by date rows between 9 preceding and current row)
from t;
我不知道SAP Hana是否支持这种语法。 SQL Server可以。
答案 1 :(得分:1)
您可以执行以下操作来满足行与行之间没有连续日期的情况。
SELECT
t1.Id
,t1.Date
,t1.Qt
,SUM(t2.Qt) as Rolling10DayQt
FROM [your-table] t1
CROSS APPLY (
SELECT *
FROM [your-same-table] a
WHERE a.Id= t1.Id
AND (a.Date > DATEADD(DAY,-10,t1.Date) AND a.Date <= t1.Date) ) t2
GROUP BY t1.Id ,t1.Date ,t1.Qt
希望有帮助。
答案 2 :(得分:0)
试试这个......
select
ID
,Date
,Qt
,SUM(Qt) OVER (ORDER BY ID ROWS BETWEEN UNBOUNDED PRECEDING AND 0 PRECEDING) AS 'Roll10Day'
from [TABLE NAME]