SQL 查询根据日期获取最新的 3 次发送，然后对记录进行分组

Question

我有一个表格，每个产品 id(pid) 都有电子邮件 sent_date 和 click_date，如下例所示：

email	pid	sent_date	click_date
A@gmail.com	1001	1 May 2021 21:12	null
A@gmail.com	1001	2 May 2021 21:12	null
A@gmail.com	1001	3 May 2021 21:12	3 May 2021 21:12
B@gmail.com	1001	1 May 2021 21:12	1 May 2021 21:12
B@gmail.com	1001	2 May 2021 21:12	2 May 2021 21:12
B@gmail.com	1001	3 May 2021 21:12	3 May 2021 21:12
A@gmail.com	1007	1 May 2021 21:12	1 May 2021 21:12
A@gmail.com	1007	2 May 2021 21:12	null
A@gmail.com	1007	3 May 2021 21:12	3 May 2021 21:12
A@gmail.com	1007	4 May 2021 21:12	null
A@gmail.com	1007	5 May 2021 21:12	null
A@gmail.com	1007	6 May 2021 21:12	null

在这个表中，我需要建立一个名为“Last 3 sent_clickCount”的新表。

所以决赛桌应该如下所示：

email	pid.	clickCount
A@gmail.com	1001	1
B@gmail.com	1001	3
A@gmail.com	1007	null

我想根据pid 考虑仅对最近 3 次发送的点击。

Answer 1

select t1.email,
       t1.pid,
       count(case when rn <= 3 then t1.click_date else null end) clickcount
  from (
select t1.email,
       t1.pid,
       row_number() over(partition by t1.email, t1.pid order by t1.sent_date desc) rn,
       t1.click_date
  from tab1 t1 
) t1
 group by t1.email,
       t1.pid
;

Answer 2

像下面这样的查询可以解决问题

with cte as (
SELECT *
,RANK() over (partition by email,pid order by sent_date desc) as ranking
FROM #temp
)
select email,pid
,sum(case when click_date is null then null else 1 end) clickcount
from cte
where ranking <=3
group by email,pid

Answer 3

你可以试试这个：

CREATE TABLE #temp 
(  email varchar(20)
  ,pid    int
  ,sent_date datetime
  ,click_date datetime
) 

INSERT INTO #temp VALUES 
('A@gmail.com',1001,'1 May 2021 21:12',null)
,('A@gmail.com',1001,'2 May 2021 21:12',null)
,('A@gmail.com',1001,'3 May 2021 21:12','3 May 2021 21:12')
,('B@gmail.com',1001,'1 May 2021 21:12','1 May 2021 21:12')
,('B@gmail.com',1001,'2 May 2021 21:12','2 May 2021 21:12')
,('B@gmail.com',1001,'3 May 2021 21:12','3 May 2021 21:12')
,('A@gmail.com',1007,'1 May 2021 21:12','1 May 2021 21:12')
,('A@gmail.com',1007,'2 May 2021 21:12',null)
,('A@gmail.com',1007,'3 May 2021 21:12','3 May 2021 21:12')
,('A@gmail.com',1007,'4 May 2021 21:12',null)
,('A@gmail.com',1007,'5 May 2021 21:12',null)
,('A@gmail.com',1007,'6 May 2021 21:12',null)


SELECT email,
       pid,
       clickCount
FROM
(
       SELECT *
              ,RANK() over (partition by email,pid order by sent_date desc) as ranking
       FROM #temp
) as ranktable
OUTER APPLY (SELECT CASE WHEN ranktable.click_date IS NOT NULL 
                         THEN COUNT(*) 
                    ELSE 
                              NULL
                    END AS clickCount 
              FROM #temp AS t 
              WHERE t.pid = ranktable.pid 
              AND t.email = ranktable.email 
              AND click_date IS NOT NULL
            ) as counttable
WHERE ranking = 1

drop table #temp

上面提到的编码应该在MSSQL上使用（SQL Server 2005以上）

首先，使用rank找出最新的3个sent_date。

其次，使用outer apply 查找点击次数。如果最新的click_date是null，则返回null，否则返回clickCount