我从PostgreSQL得到“聚合函数调用无法嵌套”错误。我尝试了不同的东西,但无法解决。
select c.*, (
select sum((count(distinct product_id))/2)
from page_views
where product_id in (c.p1, c.p2)
group by user_id, session_id
having count(distinct product_id) > 1
) freq
from (
select a.product_id p1, b.product_id p2
from (select distinct product_id from page_views) a,
(select distinct product_id from page_views ) b
where a.product_id <> b.product_id
) c ;
谢谢!
答案 0 :(得分:1)
如果您想统计在一次感受中看过两个页面的用户,那么这就是查询:
select v1.product_id, v2.product_id, count(distinct v2.user_id)
from page_views v1 join
page_views v2
on v1.user_id = v2.user_id and v1.session_id = v2.session_id and
v1.product_id < v2.product_id
group by v1.product_id, v2.product_id;
这是我能想象到的最明智的解释。
答案 1 :(得分:0)
您可以使用subselect来获取嵌套聚合函数,如下所示:
select c.*, (SELECT sum(count_column) FROM (
select (count(distinct product_id))/2 AS count_column
from page_views
where product_id in (c.p1, c.p2)
group by user_id, session_id
having count(distinct product_id) > 1
) sub_q
) freq
from (
select a.product_id p1, b.product_id p2
from (select distinct product_id from page_views) a,
(select distinct product_id from page_views ) b
where a.product_id <> b.product_id
) c ;
答案 2 :(得分:0)
我不完全理解您在此示例中要做什么,但是我想展示一个示例,说明如何在PostgresQL中使用多个聚合函数。
假设我的目标是找到max(time)-min(time)的最大值:
select max(a.trip_time) from
( select trip_id, max(time) - min(time) as trip_time
from gps_data
where date = '2019-11-16'
group by trip_id) as a;
我希望这很清楚!