每个月底的总和值

Question

我有两张桌子。

具有 CustomerID、InternalID 和 AsOfDate 的表 A。

具有 InternalID、Spend、StoreNum 和 AsOfDate 的表 B。

我希望创建一个输出，以便我获得过去 12 个月每月最后一天的每个 CustomerID 的支出总和。一个月的最后一天不是 30 或 31，它是按工作日计算的，所以我想要一个查询，查找每个月内的 MAX 日期以确定最后一天。

带有数据和查询的示例表。

Declare @MAXDATE DATETIME
Select @MAXDATE = MAX (as_of_date) from table2 

select
  s.customerID,
  sum(case when s.asofdate = CAST('11/29/21' AS DATETIME) then hn.spend end) as val_prevmonth,
  sum(case when s.asofdate = @MAXDATE then hn.spend end) as val_curr
from table1 s
join table2 hn on hn.asofdate = s.asofdate 
                 and hn.internalID = s.internalID
where 
and  (s.asofdate = CAST('11/30/21' AS DATETIME) or s.asofdate = @MAXDATE)
group by s.customerID
order by s.customerID;

表 A

CustomerID	InternalID	AsOfDate
610	016	11/30/2021
610	016	11/29/2021
610	016	11/28/2021
610	016	10/29/2021
610	016	10/15/2021
610	016	9/15/2021
610	016	9/14/2021
610	016	8/10/2021
725	527	11/30/2021
725	527	11/29/2021
725	527	11/28/2021
725	527	10/29/2021
725	527	10/15/2021
725	527	9/15/2021
725	527	9/14/2021
725	527	8/10/2021

表 B

Spend	InternalID	StoreNum	AsOfDate
5000	016	123	11/30/2021
4000	016	321	11/30/2021
3000	016	456	11/28/2021
4500	016	654	10/29/2021
5000	016	789	10/15/2021
4000	016	756	9/15/2021
9000	016	987	9/15/2021
1000	016	545	8/10/2021
6000	527	124	11/30/2021
4500	527	354	11/30/2021
9000	527	745	11/28/2021
8500	527	846	10/29/2021
7500	527	646	10/29/2021
6000	527	311	9/14/2021
3000	527	211	9/14/2021
2500	527	121	8/10/2021

表 C。注意 CustomerID 725 从 9 月开始丢失，因为他们在 9 月的最后一个工作日（即 15 日）没有进行任何交易。列名也不需要包含日期，为了清楚起见，我把它放在那里。

CustomerID	11/30/2021 TotalSpend	10/29/2021 TotalSpend	9/15/2021 TotalSpend	8/10/2021 TotalSpend
610	9000	4500	1300	1000
725	10500	16000		2500

Answer 1

您可以在 MS Sql Server 中以年月为轴。

SELECT *
FROM (
  SELECT 
    CustomerID
  , FORMAT(MAX(AsOfDate), 'yyyy-MM')+' Total Spend' AS Col
  , SUM(Spend) AS TotalSpend
  FROM
  (
    SELECT 
      t1.*
    , t2.Spend
    , t2.StoreNum
    , rnk = DENSE_RANK() OVER (PARTITION BY t1.InternalID, EOMONTH(t1.AsOfDate) ORDER BY t1.AsOfDate DESC)
    FROM Table1 t1
    JOIN Table2 t2
      ON t2.InternalID = t1.InternalID
     AND t2.AsOfDate = t1.AsOfDate
    WHERE t1.AsOfDate >= DATEADD(year,-1,DATEADD(month,-1,DATEADD(day,1,EOMONTH(GETDATE()))))
      AND t2.AsOfDate >= DATEADD(year,-1,DATEADD(month,-1,DATEADD(day,1,EOMONTH(GETDATE()))))
  ) q
  WHERE rnk = 1
  GROUP BY CustomerID, EOMONTH(AsOfDate)
) Src
PIVOT (
  SUM(TotalSpend)
  FOR Col IN (
    [2021-11 Total Spend]
  , [2021-10 Total Spend]
  , [2021-09 Total Spend]
  , [2021-08 Total Spend]
  )
) Pvt
ORDER BY CustomerID;

Answer 2

已更新 - 我已将答案修改为目标 SQL Server。

我们首先获取我们正在查看的期间的 CustomerID 列表 -

SELECT DISTINCT CustomerID
FROM TableA
WHERE AsOfDate > NOW() - INTERVAL 1 YEAR

我们会做类似的事情来获取过去一年该月最后一个交易日的列表，然后应用 ROW_NUMBER()，这样我们就有了从 1 到 12 的唯一数字来处理数据透视表。

SELECT ROW_NUMBER() OVER(ORDER BY MAX(AsOfDate) DESC) AS RowNum, MAX(AsOfDate) EndOfMonth
FROM TableA
WHERE AsOfDate > NOW() - INTERVAL 1 YEAR
GROUP BY YEAR(AsOfDate), MONTH(AsOfDate)
ORDER BY EndOfMonth ASC

然后，我们将上述查询作为两个派生表与一个 CROSS JOIN 组合在一起，以提供笛卡尔积，然后将左连接与 TableA 和 TableB。

WITH customers (CustomerID) AS (
    SELECT DISTINCT CustomerID
    FROM TableA
    WHERE AsOfDate > DATEADD(year, -1, GETDATE())
), months (RowNum, EndOfMonth) AS (
    SELECT ROW_NUMBER() OVER(ORDER BY MAX(AsOfDate) DESC) AS RowNum, MAX(AsOfDate) EndOfMonth
    FROM TableA
    WHERE AsOfDate > DATEADD(year, -1, GETDATE())
    GROUP BY YEAR(AsOfDate), MONTH(AsOfDate)
)
SELECT
    customers.CustomerID,
    SUM(CASE WHEN months.RowNum =  1 THEN TableB.Spend END) AS TotalSpend1,
    SUM(CASE WHEN months.RowNum =  2 THEN TableB.Spend END) AS TotalSpend2,
    SUM(CASE WHEN months.RowNum =  3 THEN TableB.Spend END) AS TotalSpend3,
    SUM(CASE WHEN months.RowNum =  4 THEN TableB.Spend END) AS TotalSpend4,
    SUM(CASE WHEN months.RowNum =  5 THEN TableB.Spend END) AS TotalSpend5,
    SUM(CASE WHEN months.RowNum =  6 THEN TableB.Spend END) AS TotalSpend6,
    SUM(CASE WHEN months.RowNum =  7 THEN TableB.Spend END) AS TotalSpend7,
    SUM(CASE WHEN months.RowNum =  8 THEN TableB.Spend END) AS TotalSpend8,
    SUM(CASE WHEN months.RowNum =  9 THEN TableB.Spend END) AS TotalSpend9,
    SUM(CASE WHEN months.RowNum = 10 THEN TableB.Spend END) AS TotalSpend10,
    SUM(CASE WHEN months.RowNum = 11 THEN TableB.Spend END) AS TotalSpend11,
    SUM(CASE WHEN months.RowNum = 12 THEN TableB.Spend END) AS TotalSpend12
FROM customers
CROSS JOIN months
LEFT JOIN TableA
    ON customers.CustomerID = TableA.CustomerID
    AND months.EndOfMonth = TableA.AsOfDate
LEFT JOIN TableB
    ON TableA.InternalID = TableB.InternalID
    AND TableA.AsOfDate = TableB.AsOfDate
GROUP BY customers.CustomerID
ORDER BY customers.CustomerID ASC;;

这是db<>fiddle

如果您想在结果中包含每个月的最后一个交易日，您可以将 SELECT 列表更改为 -

SELECT
    customers.CustomerID,
    MAX(CASE WHEN months.RowNum =  1 THEN months.EndOfMonth END) AS `Month1`,
    SUM(CASE WHEN months.RowNum =  1 THEN TableB.Spend END) AS `TotalSpend1`,
    MAX(CASE WHEN months.RowNum =  2 THEN months.EndOfMonth END) AS `Month2`,
    SUM(CASE WHEN months.RowNum =  2 THEN TableB.Spend END) AS `TotalSpend2`,
    ...