Question

I have a table that looks like this:

UID DATE GROUP VALUE

The data types are:

 VARCHAR DATE VARCHAR NUMERIC

Example:

abc1000 2012-09-01 1205 1000.0000
abc1000 2012-09-01 1210 1010.0000
abc1000 2012-09-02 1205 1100.0000
abc1000 2012-09-02 1210 1020.0000
def1010 2012-09-01 1205 2000.0000

I need to find the top N(15) values (VALUE) for each unique “UID+GROUP” and AVERAGE them so the output looks like this:

abc1000 1205 1050.0000
abc1000 1210 1015.0000
def1010 1205 2000.0000

Answer 1

select uid, [group], avg(value)
from
(
 select *,
        row_number() over (partition by uid, [group] order by value desc) rn 
        from yourtable
) v
where rn<=15
group by uid, [group]