我有一个包含用户项目的表格。每个用户可以具有多种类型的项目,并且可以使每个项目不止一次。我想看看每个用户拥有的每种类型的项目数量。所以我使用以下查询:
select user_name, count(item_name) as "count_item", item_name
from my_table
group by user_name, item_name
order by user_name, count_item desc;
所以我得到这样的东西:
user_name | count_item | item_name
----------+-------------+-----------
User 1 | 10 | item X
User 1 | 8 | item Y
User 2 | 15 | item A
User 2 | 13 | item B
User 2 | 7 | item C
User 2 | 2 | item X
等
现在,我想只看到每个用户的前3项。在上面的示例中,对于用户1,我想要查看项目X和Y,对于用户2,我想要查看项目A,B和C.
我怎么能这样呢?
谢谢!
答案 0 :(得分:3)
使用PARTITION BY。这样的事情应该有效:
select user_name, count_item, item_name
from (select user_name, count(item_name) as "count_item", item_name
row_number() over (partition by user_name order by count_item desc)
from my_table)
where row_number < 4
group by user_name, item_name
order by user_name, count_item desc;
答案 1 :(得分:0)
不幸的是,我没有Postgres来测试这个,但是如下所示的查询可以让你得到你想要的结果。
select user_name, item_name, item_count from
(
select user_name, item_name, count(item_name) as item_count,
dense_rank() over (order by count(item_name) desc) as item_rank
from my_table
group by user_name, item_name
)
where item_rank <= 3;