假设你有一个如下的简单表:
+---------+---------------------+
| user_id | activity_date |
+---------+---------------------+
| 23672 | 2011-04-26 14:53:02 |
| 15021 | 2011-04-26 14:52:20 |
| 15021 | 2011-04-26 14:52:09 |
| 15021 | 2011-04-26 14:51:51 |
| 15021 | 2011-04-26 14:51:38 |
| 6241 | 2011-04-26 14:51:12 |
| 168 | 2011-04-26 14:51:12 |
...
如何选择过去4周内每周至少出现一次的user_ids集合?
答案 0 :(得分:4)
select user_id,
sum(if(activity_date between now() - interval 1 week and now(),1,0)) as this_week,
sum(if(activity_date between now() - interval 2 week and now() - interval 1 week,1,0)) as last_week,
sum(if(activity_date between now() - interval 3 week and now() - interval 2 week,1,0)) as two_week_ago,
sum(if(activity_date between now() - interval 4 week and now() - interval 3 week,1,0)) as three_week_ago
from activities
where activity_date >= now() - interval 4 week
group by user_id
having this_week > 0 and last_week > 0 and two_week_ago > 0 and three_week_ago > 0
答案 1 :(得分:0)
这使用了Oracle语法,因此MySql可能需要进行一些清理,但是如果在4周内每次发生活动,您可以使用一系列子查询从表中进行选择,然后按用户汇总总计,然后只选择有活动>的用户每周0次。
select user_id from
{
select user_id, SUM(in_week_0) sum_week_0,SUM(in_week_1) sum_week_1,SUM(in_week_2) sum_week_2,SUM(in_week_3) sum_week_3 from
(
select user_id,
activity_date,
CASE
WHEN activity_date < '01-MAY-11' AND activity_date >= '24-APR-11' THEN 1
ELSE 0
END in_week_0,
CASE
WHEN activity_date < '24-APR-11' AND activity_date >= '17-APR-11' THEN 1
ELSE 0
END in_week_1,
CASE
WHEN activity_date < '17-APR-11' AND activity_date >= '10-APR-11' THEN 1
ELSE 0
END in_week_2,
CASE
WHEN activity_date < '10-APR-11' AND activity_date >= '03-APR-11' THEN 1
ELSE 0
END in_week_3
from table
)
group by user_id
)
where sum_week_0 > 0 and sum_week_1 > 0 and sum_week_2 > 0 and sum_week_3 > 0;