SQL-按分组到单独的列中

使用SQLServer2008R2。我不确定这是否可行，但我想按两列分组，将它们的差异放入新的列中。用一个例子来解释可能更容易：

BEFORE:: [DATE] [ID] [AMT] [TYPE] 2013-01-11 36374DCD-47FE-48D8-8E70-8D3B37385311 2 20 2013-01-11 36374DCD-47FE-48D8-8E70-8D3B37385311 10 21 2013-01-11 4434E2D5-1D08-45FA-AADF-F653BF9A0D97 4 20 2013-01-11 4434E2D5-1D08-45FA-AADF-F653BF9A0D97 4 21 AFTER:: [DATE] [ID] [AMT 20] [AMT 21] 2013-01-11 36374DCD-47FE-48D8-8E70-8D3B37385311 2 10 2013-01-11 4434E2D5-1D08-45FA-AADF-F653BF9A0D97 4 4

---

感谢您的帮助，谢谢

您可以使用带有

大小写的聚合函数将行转换为列：

SELECT CAST(TransDate AS DATE) AS [TransDate],
  ItemID,
  count(case when TransactionTypeID=20 then TransactionTypeID end) Amt_20,
  count(case when TransactionTypeID=21 then TransactionTypeID end) Amt_21
FROM Transactions
WHERE 
    TransDate BETWEEN '2013-01-01 10:00:00' AND '2013-02-01 10:00:00'
    AND TransactionTypeID IN (20,21)
GROUP BY CAST(TransDate AS DATE),ItemID;

由于您使用的是SQL Server，因此也可以使用
PIVOT
函数编写：

select TransDate,
  ItemId,
  [20] as Amt_20,
  [21] as Amt_21
FROM
(
  SELECT CAST(TransDate AS DATE) AS [TransDate],
    ItemID,
    TransactionTypeID
  FROM Transactions
  WHERE TransDate BETWEEN '2013-01-01 10:00:00' AND '2013-02-01 10:00:00'
    AND TransactionTypeID IN (20,21)
) d
pivot
(
  count(TransactionTypeID)
  for TransactionTypeID in ([20], [21])
) piv

bluefeet-answer可能是最好的方法，但是如果您的查询变得更复杂，您可以根据需要使用任意多的联合：）

看起来您正试图
pivot
结果。从不知道枢轴功能。非常好，谢谢！
select TransDate,
  ItemId,
  [20] as Amt_20,
  [21] as Amt_21
FROM
(
  SELECT CAST(TransDate AS DATE) AS [TransDate],
    ItemID,
    TransactionTypeID
  FROM Transactions
  WHERE TransDate BETWEEN '2013-01-01 10:00:00' AND '2013-02-01 10:00:00'
    AND TransactionTypeID IN (20,21)
) d
pivot
(
  count(TransactionTypeID)
  for TransactionTypeID in ([20], [21])
) piv

    SELECT 
        CAST(TransDate AS DATE) AS [TransDate],ItemID,COUNT(TransactionTypeID) AS [TransAmt], TransactionTypeID
    FROM 
        Transactions
    WHERE 
        TransDate BETWEEN '2013-01-01 10:00:00' AND '2013-02-01 10:00:00'
        AND TransactionTypeID IN (20)
    GROUP BY
        CAST(TransDate AS DATE),ItemID,TransactionTypeID
UNION
SELECT 
    CAST(TransDate AS DATE) AS [TransDate],ItemID,COUNT(TransactionTypeID) AS [TransAmt], TransactionTypeID
FROM 
    Transactions
WHERE 
    TransDate BETWEEN '2013-01-01 10:00:00' AND '2013-02-01 10:00:00'
    AND TransactionTypeID IN (21)
GROUP BY
    CAST(TransDate AS DATE),ItemID,TransactionTypeID