行号分区

Question

I'm using SQL Server Management Studio 2012.

Sorry for the potentially misleading title wasn't sure what to call it.

I need to workout how to find Incidents where only one Person has been mobilised to it. I figured I'd be able to achieve this by using Row_Number partition and then filtering on 1

 RowNumber    Incident MobilisedDateTime     Name
    1          abc     2019-02-08            Fred
    2          abc     2019-02-08            Roger
    3          abc     2019-02-08            Brian
    4          abc     2019-02-08            John
    1          def     2019-02-08            Joe
    1          ghi     2019-02-07            Robert
    1          jkl     2019-02-06            Jimmy
    2          jkl     2019-02-06            John

This is what happens when filtering on row number 1. This is not what I'm after because Incident 'abc' and 'jkl' had more than one person mobilised to it.

    RowNumber Incident MobilisedDateTime     Name
    1          abc     2019-02-08            Fred
    1          def     2019-02-08            Joe
    1          ghi     2019-02-07            Robert
    1          jkl     2019-02-06            Jimmy

The result I would like to see is this

    RowNumber Incident MobilisedDateTime     Name
     1          def     2019-02-08           Joe
     1          ghi     2019-02-07           Robert

Answer 1

The best method is probably not exists . Window functions are not needed:

select t.*
from t
where not exists (select 1 
                  from t t2
                  where t2.incident = t.incident and
                        t2.name <> t.name
                 );

For performance, you want an index on (incident, name) .

Answer 2

您可以尝试此查询，这将返回您的结果

select RowNumber,Incident,MobilisedDateTime,max(Name) as name,count(distinct name) as cnt from table_name group by RowNumber,Incident,MobilisedDateTime  having(count(distinct name) = 1)