Question

我有一个表，该表由三列组成-UPC，ATTRIBUTE，STORE_NUM。我有10家商店和2家UPC，每家都有不同的ATTRIBUTE。

每个商店都具有X或Y属性。我按UPC和ATTRIBUTE分组，并获取商店计数。

SELECT [UPC], [ATTRIBUTE], COUNT([STORE_NUM]) AS [COUNT]
FROM TABLEA
GROUP BY [UPC], [ATTRIBUTE]

赞这个：

UPC    ATTRIBUTE   COUNT
1      X           8
1      Y           2
2      X           1
2      Y           9

我想选择具有最高计数的UPC和ATTRIBUTE。我想要的输出是这样：

UPC      ATTRIBUTE
1        X
2        Y

我不知道如何达到期望的结果。

Answer 1

使用row_number和一个子查询：

SELECT UPC, ATTRIBUTE
FROM (
      SELECT  UPC, ATTRIBUTE, ROW_NUMBER() OVER (PARTITION BY UPC ORDER BY a_count DESC) as rn
      FROM ( SELECT [UPC],[ATTRIBUTE],COUNT([STORE_NUM]) AS [a_COUNT]
             FROM TABLEA
             GROUP BY [UPC],[ATTRIBUTE]
           ) t
     ) q
WHERE q.rn = 1

Answer 2

您可以将窗口函数用于聚合：

SELECT *
FROM (SELECT [UPC], [ATTRIBUTE], COUNT(*) AS [COUNT],
             ROW_NUMBER() OVER (PARTITION BY UPC ORDER BY COUNT(*) DESC) as seqnum
      FROM TABLEA
      GROUP BY [UPC], [ATTRIBUTE]
     ) x
WHERE seqnum = 1;

如果需要平局，请使用RANK()。

选择按列分组的两个，但仅选择具有最高COUNT（）的行

2 个答案: