Question

我正在寻找一种干净，简洁的方式来有条件地执行SUM操作。我知道这要相当直接，但我不确定如何继续。为了表明我的意思，请看一下......

    DECLARE @Test TABLE
(
    LevelID int,
    DataFieldID INT,
    RangeID INT,
    Value FLOAT
)

INSERT INTO @Test VALUES (22, 6, 117, 100)
INSERT INTO @Test VALUES (22, 6, 122, 100)
INSERT INTO @Test VALUES (22, 6, 126, 100)

INSERT INTO @Test VALUES (22, 7, 117, 100)
INSERT INTO @Test VALUES (22, 7, 122, 100)
INSERT INTO @Test VALUES (22, 7, 126, 100)

INSERT INTO @Test VALUES (23, 6, 117, 100)
INSERT INTO @Test VALUES (23, 6, 122, 100)
INSERT INTO @Test VALUES (23, 6, 126, 100)

INSERT INTO @Test VALUES (23, 7, 117, 100)
INSERT INTO @Test VALUES (23, 7, 122, 100)
INSERT INTO @Test VALUES (23, 7, 126, 100)


SELECT
    LevelID, 
    RangeID, 
    SUM(Value[where DataFieldID = 6]) / SUM(Value[where DataFieldID = 7])
FROM
    @Test
GROUP BY LevelID, RangeID

知道实现上述SUM操作的最佳方法是什么？

Answer 1

您可以在聚合中使用CASE WHEN：

SELECT
    LevelID, 
    RangeID, 
    SUM(CASE WHEN DataFieldID = 6 THEN Value ELSE 0 END) / 
    SUM(CASE WHEN DataFieldID = 7 THEN Value ELSE 0 END)
FROM
    @Test
GROUP BY LevelID, RangeID

Answer 2

我还建议使用与@Matt suggested相同的CASE，但我想你可能想要所有DataFieldId的总数，如......？也许我误解了，如果是这样的话，我会删除......

SELECT  LevelID ,
        RangeID ,
        a.total
FROM    @Test
        INNER JOIN ( SELECT SUM(Value) total ,
                            DataFieldID
                     FROM   @Test
                     GROUP BY DataFieldID
                   ) a ON [@Test].DataFieldID = a.DataFieldID
GROUP BY LevelID ,
        RangeID ,
        a.total

T-SQL条件SUM操作

2 个答案: