SQL 带 COUNT DISTINCT 的 case 语句分组

Question

我在 SQL Server 2012 中的 table 如下所示：

SaleDate    CustomerID    Customer    
2014-03-01     NULL        George    
2014-03-01     NULL        John        
2014-03-01     125         Emy    
2014-04-01     126         Natasha    
2014-04-01     127         NULL   
2014-05-01     128         Jack

我想找出每个销售日期的独立客户数量，考虑到：

有些情况下 CustomerID 丢失（但 Customer 名称已知）
并且存在 CustomerID 存在的情况（但缺少 Customer 名称）

期望的输出是：

SaleDate       UniqueCustomers
2014-03-01          3
2014-04-01          2
2014-05-01          1

我试过这个查询，但是 - 当然 - 我收到以下错误：不能在用于 GROUP BY 子句的分组依据列表的表达式中使用聚合或子查询:

SELECT SaleDate,
       CASE
           WHEN Customer IS NULL THEN COUNT(DISTINCT CustomerID)
           ELSE COUNT(DISTINCT Customer)
       END AS "UniqueCustomers"
FROM TableA
GROUP BY SaleDate,
         CASE
             WHEN Customer IS NULL THEN COUNT(DISTINCT CustomerID)
             ELSE COUNT(DISTINCT Customer)
         END

无法更新 table 以删除 NULL 值。

Answer 1

这不是正确的做法。在 case statement 之外使用 Count。

考虑到 CustomerID 是一个 Integer 列。试试这个

SELECT SaleDate,
       Count(DISTINCT CASE
                        WHEN Customer IS NULL THEN CONVERT(VARCHAR(50), CustomerID)
                        ELSE Customer
                      END) AS UniqueCustomers
FROM   Yourtable
GROUP  BY SaleDate

Answer 2

试试这个：

SELECT PreparedTable.SaleDate, SUM(PreparedTable.Uniques)
FROM
    (
    SELECT SaleDate, COUNT(DISTINCT CustomerID) as Uniques
        FROM TableA
        GROUP BY SaleDate
    UNION ALL
    SELECT SaleDate, COUNT(DISTINCT Customer)
        FROM TableA
        WHERE CustomerID is null
        GROUP BY SaleDate
    ) PreparedTable

GROUP BY SaleDate

SQL 带 COUNT DISTINCT 的 case 语句分组

SQL group by case statement with COUNT DISTINCT

sql

sql-server-2012