Question

我有一个表格，其重复数据类似于以下示例：

ID | ACCNO  | ACCNAME    | ADDRESS1     | ADDRESS2    | City
1  | 1001   | Joe B Ltd  | 123 Street1  |             | London
2  | 1001   | JoeB Ltd   | 123 Street1  |             | London
3  | 1001   | JoeB Ltd   | 123 Street1  |             | London
4  | 1001   | JoeB Ltd   | 123 Street1  | London      | London
5  | 1001   | JoeB Ltd   | 129 Street9  |             | London

ID当前是唯一的主键，但是当重复删除时，ACCNO应该是。

我见过很多查询要删除重复记录，例如https://stackoverflow.com/a/18719814/4949859

但是我想根据重复行的数量选择要保留的行。我相信如果我从具有最高计数的分组项目中选择一行，我最有可能获得格式正确的地址。

在我的例子中使用“NOT IN（SELECT MAX”或“MIN”会在我的情况下留下错误的记录。

但是，当我使用GROUP BY获得最高计数时，我不能包含ID字段。

SELECT COUNT(ID), ACCNO, ACCNAME, ADDRESS1, ADDRESS2, CITY FROM SUPPLIERS GROUP BY ACCNO, ACCNAME, ADDRESS1, ADDRESS2, CITY ORDER BY COUNT(ID) DESC

这会得到结果：

Count(ID) | ACCNO  | ACCNAME    | ADDRESS1     | ADDRESS2    | City
2         | 1001   | JoeB Ltd   | 123 Street1  |             | London
1         | 1001   | Joe B Ltd  | 123 Street1  |             | London
1         | 1001   | JoeB Ltd   | 123 Street1  | London      | London
1         | 1001   | JoeB Ltd   | 129 Street9  |             | London

希望我有意义。我不知道如何从计数最高的组返回ID（任何）。有没有人知道我怎么做到这一点？

编辑：

我上面的示例将除ID以外的所有列分组并获取计数，第2行和第3行将组合在一起，组计数为2（其余的计数ID为1，因为它们都是唯一的）所以我想要保持第2行或第3行，无论哪一个都是相同的。

编辑2：

我认为这会起作用：

DELETE
FROM SUPPLIERS
WHERE ID NOT IN 
 (SELECT TOP 1 MAX(ID) FROM SUPPLIERS  
  Group By ACCNO, ACCNAME, ADDRESS1, ADDRESS2, CITY 
  ORDER BY COUNT(ID) DESC)

不幸的是，这会删除除一条记录以外的所有记录，其选择版本看起来很有希望：

SELECT *
FROM SUPPLIERS a
WHERE ID NOT IN 
 (SELECT TOP 1 MAX(ID) FROM SUPPLIERS b 
  WHERE a.ACCNO = b.ACCNO Group By ACCNO, ACCNAME, ADDRESS1, ADDRESS2, CITY 
  ORDER BY COUNT(ID) DESC)

答案：

感谢用户1751825（标记为答案让我最接近最终结果）

DELETE FROM SUPPLIERS WHERE ID IN (SELECT ID
FROM SUPPLIERS a
 WHERE ID NOT IN 
  (SELECT TOP 1 MAX(ID) FROM SUPPLIERS b 
  WHERE a.ACCNO = b.ACCNO Group By ACCNO, ACCNAME, ADDRESS1, ADDRESS2, CITY 
  ORDER BY COUNT(ID) DESC))

Answer 1

根据我的理解，在您提供的示例中，您希望保留记录ID = 5并删除其余记录。

WITH CTE AS(
   SELECT ID, ACCNO, ACCNAME, ADDRESS1, ADDRESS2, CITY,
       RN = ROW_NUMBER()OVER(PARTITION BY ACCNO ORDER BY ID DESC)
  FROM SUPPLIERS
)
DELETE FROM CTE WHERE RN > 1

这应该可以做到！

Answer 2

我认为这应该做你需要的。

delete from SUPPLIERS
where ID NOT IN (
    Select max(ID)
    FROM SUPPLIERS
    Group by ACCNO
)

Answer 3

这对我有用。

表格

ID ACCNO ACCNAME ADDRESS1 ADDRESS2 City 1 1001 Joe B Ltd 123 Street1 London 2 1001 JoeB Ltd 123 Street1 London 3 1001 JoeB Ltd 123 Street1 London 4 1001 JoeB Ltd 123 Street1 London London 5 1001 JoeB Ltd 129 Street9 London 6 67 Nise Gata1 7 67 Nisse Gata2 8 67 Nisse Gata1 Haninge Stockholm

<强> RESULT ：

ACCNO ACCNAME ADDRESS1 ADDRESS2 City 1001 JoeB Ltd 123 Street1 London London 67 Nisse Gata1 Haninge Stockholm

<强>代码：

select distinct [ ACCNO ], FIRST_VALUE([ ACCNAME ]) OVER (PARTITION BY [ ACCNO ] ORDER BY case when [ ACCNAME ] is null then 1 else 0 end, rownumber ) as [ ACCNAME ], FIRST_VALUE([ ADDRESS1 ]) OVER (PARTITION BY [ ACCNO ] ORDER BY case when [ ADDRESS1 ] is null then 1 else 0 end, rownumber ) as [ ADDRESS1 ], FIRST_VALUE([ ADDRESS2 ]) OVER (PARTITION BY [ ACCNO ] ORDER BY case when [ ADDRESS2 ] is null then 1 else 0 end, rownumber ) as [ ADDRESS2 ], FIRST_VALUE([ City]) OVER (PARTITION BY [ ACCNO ] ORDER BY case when [ City] is null then 1 else 0 end, rownumber ) as [ City] FROM ( SELECT [ ACCNO ] ,[ ACCNAME ] ,[ ADDRESS1 ] ,case when ltrim(rtrim([ ADDRESS2 ] )) = '' then null else [ ADDRESS2 ] end as [ ADDRESS2 ] -- spaces = NULL ,case when ltrim(rtrim([ City] )) = '' then null else [ City] end as [ City] ,count(*) as quantity ,ROW_NUMBER() OVER ( PARTITION BY [ ACCNO ] ORDER BY [ ACCNO ], count(*) desc ) as rownumber FROM [dbo].[test_sql] GROUP BY cube ( [ ACCNO ] ,[ ACCNAME ] ,[ ADDRESS1 ] ,[ ADDRESS2 ] ,[ City]) HAVING [ ACCNO ] is not null ) myGroup

SQL删除重复项 - 保持记录包含在最高计数中

3 个答案: