T-SQL中对PIVOT表的分区

Question

我喜欢使用窗口函数，因此我使用了简单的row_number（）函数根据employeeID拆分一些数据。我写的东西行得通，但我不禁想到可以使用PIVOT编写。这是代码：-

WITH [Roles] AS
(
    SELECT a.EmployeeID, a.StaffRole,
    ROW_NUMBER() OVER (PARTITION BY a.EmployeeID ORDER BY a.StaffRole) AS RowNum
     FROM [GCHHS_Reporting].[dbo].[CathLab_Staff] a
)
SELECT a.EmployeeID, b1.StaffRole AS SR#1, b2.StaffRole AS SR#2, b3.StaffRole AS SR#3, b4.StaffRole AS SR#4, b5.StaffRole AS SR#5 FROM [GCHHS_Reporting].[dbo].[CathLab_Staff] a
    INNER JOIN [Roles] b1
        ON a.EmployeeID = b1.EmployeeID AND b1.RowNum = 1
    LEFT JOIN [Roles] b2
        ON a.EmployeeID = b2.EmployeeID AND b2.RowNum = 2
    LEFT JOIN [Roles] b3
        ON a.EmployeeID = b3.EmployeeID AND b3.RowNum = 3
    LEFT JOIN [Roles] b4
        ON a.EmployeeID = b4.EmployeeID AND b4.RowNum = 4
    LEFT JOIN [Roles] b5
        ON a.EmployeeID = b5.EmployeeID AND b5.RowNum = 5

上面的代码显示自开始以来每个员工所担任的角色。所以，我的问题是；有PIVOT这样的方法吗？

以下是原始数据的示例：-

以下是窗口功能数据的示例：-

任何想法或我是如何做到的，最好的方法？

Answer 1

是的，这比对表进行5次扫描要容易得多。与使用PIVOT相比，我发现一些条件聚合要容易得多：

WITH [Roles] AS(
    SELECT CLS.EmployeeID,
           CLS.StaffRole,
           ROW_NUMBER() OVER (PARTITION BY CLS.EmployeeID ORDER BY CLS.StaffRole) AS RowNum
    FROM [GCHHS_Reporting].[dbo].[CathLab_Staff] CLS --a? CathLab doesn't start with "a".
)
SELECT R.EmployeeID,
       MAX(CASE R.RowNum WHEN 1 THEN R.StaffRole END) AS SR1,
       MAX(CASE R.RowNum WHEN 2 THEN R.StaffRole END) AS SR2,
       MAX(CASE R.RowNum WHEN 3 THEN R.StaffRole END) AS SR3,
       MAX(CASE R.RowNum WHEN 4 THEN R.StaffRole END) AS SR4,
       MAX(CASE R.RowNum WHEN 5 THEN R.StaffRole END) AS SR5
FROM Roles R --nor does Roles begin with a "b"
GROUP BY R.EmployeeID;

我无法针对您的数据对此进行测试（因为它是图像），但是我做了一个小样本，返回的结果是在之后。 db<>fiddle

Answer 2

如果要在查询中计算一些度量，则

PIVOT确实很有意义。在您的简单示例中，Larnu提到一个case语句就足够了。