聚合查询中的非分组列

Question

说我有一个有三列的表：c1，c2和c3。我想在每个c1中得到c2的最大值和包含最大值的行的c3的值。

这是一个例子。说MyTable包含以下记录：

c1   c2   c3
---  ---  ---
1    32   12
1    27   15
1    12   21
2    23   33
2    36   49
3    48   17
3    21   50
3    7    25

然后我的查询应该返回

c1   MAX(c2)  c3
---  -------  ---
1    32       12
2    36       49
3    48       17

但当我查询

SELECT c1, MAX(c2), c3 FROM MyTable GROUP BY c1

我收到错误：

Column 'MyTable.c3' is invalid in the select list because it is not 
contained in either an aggregate function or the GROUP BY clause.

我该如何修复此查询？我尝试过子查询，连接等无济于事。

Answer 1

你可以利用CASE和OVER()条款。

DECLARE 
    @MyTable TABLE(c1 INT, c2 INT, c3 INT)

INSERT INTO @MyTable 
VALUES
(1,    32 ,  12),
(1,    27 ,  15),
(1,    12 ,  21),
(2,    23 ,  33),
(2,    36 ,  49),
(3,    48 ,  17),
(3,    21 ,  50),
(3,    7  ,  25)


SELECT *
FROM (
    SELECT   
        c1
    ,   MAX(c2) OVER(PARTITION BY c1) c2
    ,   CASE WHEN c2 = MAX(c2) OVER(PARTITION BY c1) THEN c3 ELSE NULL END c3
    FROM @MyTable
) D
WHERE 
    c3 IS NOT NULL 

Answer 2

我认为where中的相关子查询是一个简单的解决方案：

select t.*
from t
where t.col2 = (select max(t2.col2) from t t2 where t2.col1 = t.col1);

这通常具有非常好的性能特征。更传统的解决方案使用row_number(：

select t.*
from (select t.*,
             row_number() over (partition by col1 order by col2 desc) as seqnum
      from t
     ) t
where seqnum = 1;

如果你有一个单独的col1值表，那么apply也可以有很好的表现：

select t.*
from table_col1 c1 cross apply
     (select top (1) t.*
      from t 
      where t.col1 = c1.col1
      order by t.col2 desc
     ) t;

Answer 3

你可以试试：

SELECT a.c1, a.c2, a.c3 
FROM
MyTable a
INNER JOIN
(SELECT c1, MAX(c2) as 'max_c2' FROM MyTable GROUP BY c1) b
ON
a.c1 = b.c1
AND
a.c2 = b.max_c2