Question

对于购买者电子邮件地址列表，我试图为每个购买者返回一行，其中包含“1stOrderDate”，“2ndOrderDate”，“3rdOrderDate”和“TotalNumberofOrders”列

我已尝试在子查询的WHERE子句中使用ROW_Number函数，但是它报告WHERE子句中不允许使用Windowed函数，因此我将非常感谢有关如何填写下面的??? s的任何帮助！

SELECT 
PT.email AS 'Email',
MIN(OT.orderdate) AS '1stOrderDate',
???               AS '2ndOrderDate',
???               AS '3rdOrderDate',
COUNT(DISTINCT OT.order_reference) AS 'TotalNumberOfOrders'    

FROM dbo.Orders AS OT
JOIN dbo.Purchaser AS PT ON OT.account_reference = PT.account_reference

GROUP BY PT.Email

Answer 1

您可以使用row_number()和条件聚合：

执行此操作

SELECT PT.email,
       MAX(CASE WHEN seqnum = 1 THEN OT.OrderDate END) as OrderDate_1,
       MAX(CASE WHEN seqnum = 2 THEN OT.OrderDate END) as OrderDate_2,
       MAX(CASE WHEN seqnum = 3 THEN OT.OrderDate END) as OrderDate_3,
       COUNT(DISTINCT OT.order_reference) AS TotalNumberOfOrders   
FROM (SELECT o.*, ROW_NUMBER() OVER (PARTITION BY account_reference ORDER BY o.orderdate) as seqnum
      FROM dbo.Orders o
     ) OT JOIN
     dbo.Purchaser PT
     ON OT.account_reference = PT.account_reference
GROUP BY PT.Email

几点说明：

不要使用单引号作为列别名。而是选择不需要转义的列名。
对于segnum = 1逻辑，您可以使用MIN()，但我认为一致性是一个好处。

编辑：

我的猜测是问题是account_reference和email之间的区别。试试这个：

SELECT email,
       MAX(CASE WHEN seqnum = 1 THEN OT.OrderDate END) as OrderDate_1,
       MAX(CASE WHEN seqnum = 2 THEN OT.OrderDate END) as OrderDate_2,
       MAX(CASE WHEN seqnum = 3 THEN OT.OrderDate END) as OrderDate_3,
       COUNT(DISTINCT OT.order_reference) AS TotalNumberOfOrders   
FROM (SELECT o.*, ROW_NUMBER() OVER (PARTITION BY pt.email ORDER BY o.orderdate) as seqnum
      FROM dbo.Orders o JOIN
           dbo.Purchaser PT
           ON OT.account_reference = PT.account_reference
     ) OT 
GROUP BY PT.Email

退还每位买家的第一，第二和第三个最早购买日期

1 个答案: