Group by count distinct

Question

mytable
id | yy
1 | 78
2 | 78
3 | 78
3 | 79
3 | 79
4 | 79
5 | 79
5 | 80
Desired output:
yy | id_count
78 | 3
79 | 2
80 | 0
Following query doesn't work, as it doesn't take into account that id was already counted
select yy, count(distinct id) as id_count
from mytable
group by yy
--output
yy | id_count
78 | 3
79 | 4
80 | 1
Hope this makes sense.
Ideas?

Answer 1

Hi,
You only want to count each id once, with the first (that is, lowest) yy: is that right?
Here's one way:
WITH     got_r_num    AS
     SELECT  id
     ,     yy
     ,     ROW_NUMBER () OVER ( PARTITION BY  id
                               ORDER BY          yy
                       ) AS r_num
     FROM    my_table
SELECT       yy
       COUNT ( CASE
                  WHEN  r_num = 1
                THEN  id
              END
          )     AS id_cnt
FROM       got_r_num
GROUP BY  yy
ORDER BY  yy
;Doing anything for the first of each id is probably a job for "ROW_NUMBER () OVER (PARTITION BY id ...)".