给定天数的累积月份

Question

我的表格如下

T1 (date int, dateYearMonth int, userID int, shopID int, amount numeric(18,2))

示例数据：

(20151212, 201512, 1, 1, 50.00),
(20151213, 201512, 1, 1, 30.00),
(20160110, 201601, 1, 1, 13.00)

我想为用户返回每个月的累计金额， 我的疑问是：

select t.dateYearMonth, t.userID,
sum(t.sum_amount) over
(partition by t.userID order by t.dateYearMonth
 rows between unbounded preceding and current row)
from (select dateYearMonth, userID, sum(amount) as sum_amount FROM T1
group by dateYearMonth, userID) t

有更优化的方法吗？ 输出预期

(201512, 1, 1, 80.00), (201601, 1, 1, 93.00)

Answer 1

您可以使用：

SELECT DISTINCT dateYearMonth, userID
       ,total = SUM(amount) OVER(PARTITION BY userID ORDER BY dateYearMonth)
FROM T1;

的 LiveDemo

Answer 2

我认为您的查询非常好。我认为它不会变得更好。

确保您有覆盖索引，如下所示：

CREATE NONCLUSTERED INDEX [IX] ON T1
(
    userID ASC
    ,dateYearMonth ASC
) INCLUDE(amount);

使用此索引，带有示例数据的临时表的执行计划如下所示：

它只读取表中的所有行（索引），并对它们进行分组/求和，无需额外排序。