如何使用 Unpivot 函数计算平均天数答案

【问题标题】：How to calculate the Average days with the Unpivot function如何使用 Unpivot 函数计算平均天数
【发布时间】：2022-11-14 05:08:40
【问题描述】：

我正在尝试计算每个工作流程的平均天数。

试试看

DROP TABLE IF EXISTS #ProcessLog;
GO

CREATE TABLE #ProcessLog
(
WorkFlow        VARCHAR(100),
ExecutionDate   DATE,
PRIMARY KEY (WorkFlow, ExecutionDate)
);
GO

INSERT INTO #ProcessLog VALUES
('Alpha','6/01/2018'),('Alpha','6/14/2018'),('Alpha','6/15/2018'),
('Bravo','6/1/2018'),('Bravo','6/2/2018'),('Bravo','6/19/2018'),
('Charlie','6/1/2018'),('Charlie','6/15/2018'),('Charlie','6/30/2018');
GO

我找到了一个使用 INNER JOIN 函数的解决方案，但我想知道是否有更简单的方法，例如 PIVOT 或 UNPIVOT 或 Window 函数？

WITH cte as ( SELECT ROW_NUMBER() OVER(ORDER by Workflow) as n, * 
              FROM       #ProcessLog as p1 )

SELECT cte.workflow, (abs(DATEDIFF(DAY, cte.ExecutionDate, cte2.ExecutionDate))
      + abs(DATEDIFF(DAY, cte2.ExecutionDate, cte3.ExecutionDate)) ) / 2 as average_days
FROM       CTE  
INNER JOIN CTE as cte2  ON cte.n +1 = cte2.n  AND cte.WorkFlow  = cte2.WorkFlow
INNER JOIN CTE as cte3  ON cte2.n +1 = cte3.n AND cte2.WorkFlow = cte3.WorkFlow

【问题讨论】：

标签： sql sql-server window-functions date-arithmetic

【解决方案1】：

您需要 CTE 中日期的 LAG 并且可以使用 DATEDIFF

DROP TABLE IF EXISTS #ProcessLog;


CREATE TABLE #ProcessLog
(
WorkFlow        VARCHAR(100),
ExecutionDate   DATE,
PRIMARY KEY (WorkFlow, ExecutionDate)
);


INSERT INTO #ProcessLog VALUES
('Alpha','6/01/2018'),('Alpha','6/14/2018'),('Alpha','6/15/2018'),
('Bravo','6/1/2018'),('Bravo','6/2/2018'),('Bravo','6/19/2018'),
('Charlie','6/1/2018'),('Charlie','6/15/2018'),('Charlie','6/30/2018');

WITH CTE As(
  SELECT
  WorkFlow
  ,ExecutionDate
  ,  LAG(ExecutionDate) OVER(PARTITION BY WorkFlow ORDER BY ExecutionDate) date2 
  FROM
  #ProcessLog)
SELECT WorkFlow,AVG(DATEDIFF(day, date2,ExecutionDate)) FROM CTE GROUP BY WorkFlow

WorkFlow	(No column name)
Alpha	7
Bravo	9
Charlie	14

【讨论】：

很好的主意。简单而简短。