Question

我正在尝试从看起来像这样的表中拉动活动订户：

StartDate                EndDate                  UpgradeProcessed    Cancelled  Recurring  Created
2018-09-05 09:46:30.000  2018-10-07 09:46:30.000  1                   0          1          2018-09-05 10:54:57.363
2018-09-05 09:26:32.000  2018-12-07 09:26:32.000  1                   0          1          2018-09-05 10:41:09.460
2018-09-05 09:17:49.000  2018-12-07 09:17:49.000  1                   0          1          2018-09-05 10:27:14.707
2018-09-05 09:21:16.000  2018-12-07 09:21:16.000  1                   0          1          2018-09-05 10:27:06.593
2018-09-05 09:23:04.000  2018-10-07 09:23:04.000  1                   0          1          2018-09-05 10:27:05.877
2018-09-04 09:09:42.000  2018-12-07 09:09:42.000  1                   0          1          2018-09-04 10:13:59.847
2018-09-04 09:02:29.000  2018-10-07 09:02:29.000  1                   0          1          2018-09-04 10:13:42.693
2018-09-04 09:27:23.000  2018-12-07 09:27:23.000  1                   0          1          2018-09-04 09:27:55.110
2018-09-04 08:09:31.000  2018-12-07 08:09:31.000  1                   0          1          2018-09-04 09:19:18.193
2018-09-04 08:12:20.000  2018-10-07 08:12:20.000  1                   0          1          2018-09-04 09:19:18.117
2018-09-04 08:12:20.000  2018-10-07 08:12:20.000  1                   0          1          2018-09-04 09:19:18.117

我想按天或周进行分组，并且每天都必须包括当天有效的所有记录，即使它们包含在先前的行中。

在上述情况下，我们将获得以下内容：

Date        Count
2018-09-05  11
2018-09-04  6

我每天可以一次提取数据，如下所示：

SELECT COUNT(*) FROM [Subscriptions]
 WHERE DATEADD(day, DATEDIFF(day,0, EndDate),0) >= DATEADD(day, DATEDIFF(day,0,Cast('2018-09-05' AS datetime)),0)
 AND DATEADD(day, DATEDIFF(day,0, Created),0) <= DATEADD(day, DATEDIFF(day,0,Cast('2018-09-05' AS datetime)),0)

我不确定如何添加GROUP BY类型子句，该子句将包括当天活跃的每一行。

Answer 1

尝试一下

SELECT CAST(StartDate AS DATE),COUNT(1) 
FROM [Subscriptions]
WHERE DATEADD(day, DATEDIFF(day,0, EndDate),0) >= DATEADD(day, DATEDIFF(day,0,Cast('2018-09-05' AS datetime)),0)
AND DATEADD(day, DATEDIFF(day,0, Created),0) <= DATEADD(day, DATEDIFF(day,0,Cast('2018-09-05' AS datetime)),0)
GROUP BY CAST(StartDate AS DATE)

Answer 2

好，找到答案了，我从here修改了一个脚本，制成了一个数字表，并使用该页面上的查询来启发灵感。

SELECT [stats.dates].thedate, COUNT(*)
FROM   [stats.dates]
JOIN   Subscriptions ON [stats.dates].thedate >= cast([StartDate] as DATE) AND [stats.dates].thedate <= cast([EndDate] as DATE)
WHERE [stats.dates].thedate >= '2017-09-04' AND [stats.dates].thedate <= '2018-09-04'
GROUP  BY [stats.dates].thedate

where子句指示要获取数据的日期范围。

MSSQL选择按日/周分组的活动订户

2 个答案: