Question

There is the table below: (columns: ID-CAUSE-WORK)

ID  | CAUSE | WORK
A   |   C1  |  W1
B   |   C1  |  W1
C   |   C1  |  W1
D   |   C1  |  W1
E   |   C1  |  W2
F   |   C1  |  W2
G   |   C1  |  W2
H   |   C1  |  W3
I   |   C1  |  W3
FF  |   C2  |  W4
FG  |   C2  |  W4
FG  |   C2  |  W1
FG  |   C2  |  W1
FG  |   C2  |  W6

I want the two max values of work’s count per cause. That is, with a simple count(work) group by cause, the result would be:

cause | work| count(work)
c1    |  w1 | 4
c1    |  w2 | 3
c1    |  w3 | 2
c2    |  w4 | 2
c2    |  w1 | 2
c2    |  w6 | 1

I want to take only the 2 max counts works per cause:

c1    |  w1 | 4
c1    |  w2 | 3
c2    |  w4 | 2
c2    |  w1 | 2

Answer 1

This should work:

select cause,
       work,
       cnt as "COUNT"
from (
  select cause, 
         work, 
         count(work) as cnt,
         row_number() over (partition by cause order by count(work) desc, work desc) as rown
  from your_table group by cause, work
) where rown <= 2;