0

我有一个老化查询，它以 12、24 和 36 个月为间隔对购买的物品数量及其成本进行老化。查询运行得很好，但是与用于查找特定供应商总数的单个查询相比，我的总数不会相加。

见下文。

    SELECT s.SupplierNumber as Supplier_Number 
    , s.suppliername as Supplier_Name
    , i.supplierpartnum as Part_No
    , SUBSTRING(e.account_code,1,5)/*+'-'+RIGHT(e.account_code,7)*/ as Account_Code
    , CASE WHEN e.reference_code = ''
        THEN 'NOREF'
        ELSE reference_code
        END as Reference_Code
    , i.Commodity
    , i.ShortDscrptn as Part_Desc
    , i.unitofmeasure as UOM
    , i.unitprice as Unit_Price
, CASE WHEN h.OrderDate >= '2012-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN COUNT(i.Quantity)
        ELSE 0 END as Annual_Qty_12
, CASE WHEN h.OrderDate >= '2011-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN COUNT(i.Quantity)
        ELSE 0 END as Annual_Qty_24 
, CASE WHEN h.OrderDate >= '2010-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN COUNT(i.Quantity)
        ELSE 0 END as Annual_Qty_36 
, CASE WHEN h.OrderDate >= '2012-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN SUM(i.UnitPrice) 
        ELSE 0 END as  Annual_Spend_12          
, CASE WHEN h.OrderDate >= '2011-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN SUM(i.UnitPrice) 
        ELSE 0 END as  Annual_Spend_24  
, CASE WHEN h.OrderDate >= '2010-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN SUM(i.UnitPrice) 
        ELSE 0 END as  Annual_Spend_36  
FROM ekp.dbo.SUPPLIER s   
JOIN ekp.dbo.ORDERHEADER h
    ON s.SupplierID = h.SupplierID
JOIN ekp.dbo.ORDERITEM i
    ON h.OrderID = i.OrderID
JOIN mgnt.dbo.ematch e
    ON h.PONumber = e.po_ctrl_num
    AND i.LineNumber = e.po_sequence_id
WHERE h.OrderDate >= '2010-01-01'
AND h.OrderDate <= '2012-12-31' 
GROUP BY s.SupplierNumber, s.SupplierName, i.SupplierPartNum,  e.account_code
, i.Commodity, i.UnitOfMeasure, i.UnitPrice, i.ShortDscrptn, h.OrderDate
, e.reference_code

我可以继续将结果复制到电子表格中，并对Annual_Spend_12特定的列求和Supplier，但它与以下内容不匹配：

    SELECT SUM(i.unitprice) as Annual_Spend_12 
    FROM ekp.dbo.ORDERITEM i
JOIN ekp.dbo.ORDERHEADER h 
    ON i.OrderID = h.OrderID
JOIN ekp.dbo.Supplier s
    ON h.SupplierID = s.SupplierID
JOIN mgnt.dbo.ematch e
    ON h.PONumber = e.po_ctrl_num
    AND i.LineNumber = e.po_sequence_id
WHERE s.SupplierNumber = '15302'
AND h.OrderDate >= '2012-01-01'
AND h.OrderDate <= '2012-12-31'

上面这个查询的总和总是比顶部查询高得多。我还对整个聚合进行了求和和计数，但它也不匹配，实际上，它产生的结果与顶部查询相同：

    , COUNT(CASE WHEN h.OrderDate >= '2012-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN i.Quantity
        ELSE 0 END) as Annual_Qty_12
, COUNT(CASE WHEN h.OrderDate >= '2011-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN i.Quantity
        ELSE 0 END) as Annual_Qty_24 
, COUNT(CASE WHEN h.OrderDate >= '2010-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN i.Quantity
        ELSE 0 END) as Annual_Qty_36 
, SUM(CASE WHEN h.OrderDate >= '2012-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN i.UnitPrice 
        ELSE 0 END) as  Annual_Spend_12         
, SUM(CASE WHEN h.OrderDate >= '2011-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN i.UnitPrice 
        ELSE 0 END) as  Annual_Spend_24 
, SUM(CASE WHEN h.OrderDate >= '2010-01-01'
        AND h.OrderDate <= '2012-12-31'
        THEN i.UnitPrice 
        ELSE 0 END) as  Annual_Spend_36

我非常害怕为每个聚合编写一个带有临时表的存储过程。

任何帮助表示赞赏。

4

1 回答 1

0

我的猜测是测试是错误的：您在 Excel 中为第一个查询添加了错误的值。以下是在 Excel 中累加值的三种有效方法：

  • 过滤值并突出显示该列。Excel 会将所选单元格中的值相加（Excel 2007 及更高版本）。
  • 使用“sumif()”或“sum({})”添加给定供应商的总数
  • 使用数据透视表

group bya的结果不一定是排序的。它们可能看起来已排序，但仍可能未排序。进行聚合的一种方法是对数据进行排序，然后挑选出结果。但是，还有其他方法，例如哈希聚合。而且，在多线程环境中，您不知道哪个线程会以什么顺序返回值。

于 2013-01-10T20:11:36.183 回答