Reputation: 810
This is referring to the question i asked before, and got a really quick answaer (max count together in an sql query). The Problemset is similar, but the solution in the prevous question would force me to access a db in a loop which will cause performance problems. So what i have now, after some joins is:
id | description
0 | bla
0 | blub
0 | bla
1 | blablub
1 | bla
... | ...
As u can see, now the id is not a primary key anymore. What i want is to get the most used description for each id in the resultset. It should look something like this:
id | most_popular_description | times_the_desc_appeared_for_an_id
0 | bla | 2
1 | blablub | 1
... | ... | ...
Upvotes: 0
Views: 1813
Reputation: 4022
I think you can use the dense_rank() analytic function to get the top N for each group set.
Something like this:
select id, description, times_the_desc_appeared_for_an_id
from
(
select id, description, count(description) times_the_desc_appeared_for_an_id
dense_rank() over (partition by id, description order by count(description) desc) position
from mytable
group by id, description
)
where
position <= 3
order by id, times_the_desc_appeared_for_an_id;
Upvotes: 0
Reputation: 728
If you only want the most popular items, then I believe this should give you the result set you're looking for. There are other ways of doing this, but stats_mode is the easiest way to obtain the "most prevalent" value in a group (i.e. the Mode).
SELECT t.id,
t.description AS most_popular_description,
COUNT(*) AS times_the_desc_appeared_for_an_id
FROM mytable t INNER JOIN (
SELECT id, stats_mode(description) AS desc FROM mytable GROUP BY id
) a ON t.id = a.id AND t.description = a.desc
GROUP BY t.id, t.description;
Note that the nested query (inline view) is necessary since you also want the count.
Upvotes: 1
Reputation: 32094
This should do the trick.
select id, description, COUNT(description)
from mytable
group by id, description
order by 3 desc
Upvotes: 1