如何根据不同的分组条件得到两个数量列的总和？

Posted 2023-02-16

【中文标题】如何根据不同的分组条件得到两个数量列的总和？

【英文标题】：How to get sum of two quantity columns based on different grouping conditions?

【发布时间】：2019-11-13 16:21:42

【问题描述】：

从 table1 和 table2 的内部连接，交货时加入，我想根据分组工厂、零件和 exp_ship_date 获得 qty_to_ship 的总和，并根据分组工厂、零件和 book_date 获得 qty_shipped 的总和。这个想法是根据 book_date 计算发货数量，根据 exp_ship_date 计算发货数量。

我尝试使用子查询但无法得到正确答案

我尝试了什么

DECLARE @REPORT AS NVARCHAR(50)='2019-06-19 00:00:00.000'
SELECT
    PLANT,
    PARTS,
    YEAR(P_DATE) AS P_DATE,MONTH(P_DATE) AS P_DATE,DAY(P_DATE) AS P_DATE,
    ES_DATE AS ES_DATE,
    SUM(CASE WHEN (P_DATE<@REPORT AND P_DATE > DATEADD(DAY,-7,@REPORT)) THEN QTY_PICKED ELSE 0 END) AS SHIPPED,
    SUM(CASE WHEN ES_DATE>@REPORT AND ES_DATE < DATEADD(DAY,7,@REPORT) THEN QTY_MII ELSE 0 END) -SUM(CASE WHEN ES_DATE>@REPORT AND ES_DATE < DATEADD(DAY,7,@REPORT) THEN QTY_PICKED ELSE 0 END)  AS TO_SHIP
      FROM TABLE1 T1
      INNER JOIN
      TABLE2 T2 
      ON
      T1.DELIVERY = T2.DELIVERY

WHERE PLANT = 'XXX'
GROUP BY
    GROUPING SETS (
        (PLANT, PARTS,YEAR(P_DATE),MONTH(P_DATE),DAY(P_DATE)),
        (PLANT, PARTS,ES_DATE)
    )
ORDER BY
PLANT,
PARTS

【问题讨论】：

请展示您的尝试以及预期结果

在“我试过的”下面添加了

【参考方案1】：

您可以使用OVER 子句根据每行的不同分组计算聚合：

SELECT *
     ,SUM(qty_to_ship) OVER (PARTITION BY plant, parts, exp_ship_date)
     ,SUM(qty_shipped) OVER (PARTITION BY plant, parts, book_date)
FROM table1 A
INNER JOIN table2 B
    ON A.[Delivery] = B.[Delivery];

然后留下您需要的列并使用DISTINCT 删除重复的行。例如：

SELECT DISTINCT A.develiery
               ,A.plant
               ,A.parts
               ,SUM(qty_to_ship) OVER (PARTITION BY plant, parts, exp_ship_date)
               ,SUM(qty_shipped) OVER (PARTITION BY plant, parts, book_date)
FROM table1 A
INNER JOIN table2 B
    ON A.[Delivery] = B.[Delivery];

【讨论】：

我收到错误 PLANT 在选择列表中无效，因为它既不包含在聚合函数或 GROUP BY 子句中。

DECLARE @REPORT AS NVARCHAR(50)='2019-06-19 00:00:00.000' SELECT *, SUM(CASE WHEN (P_DATE DATEADD(DAY,-7 ,@REPORT)) THEN QTY_PICKED ELSE 0 END) OVER(PARTITION by Plant,Parts,P_DATE)AS SHIPPED, SUM(CASE WHEN E_DATE>@REPORT AND E_DATE @REPORT AND E_DATE

您需要为每个聚合函数应用OVER 子句。