问题引入
最近经常碰到这样的问题,每天每个城市播放最多的10首歌,某月每支股票连续下跌/上涨的最大天数,用户连续活跃的最大天数,初步看起来都和分析函数相关,考验逻辑思维和写复杂SQL的能力。
以用户连续活跃的最大天数为例
以Oracle的分析函数语法说明,首先模拟一些用户活跃的数据,
-- 建表语句
DROP TABLE sigin;
create table sigin(
userid int,
sigindate varchar2(20)
);
-- 模拟数据插入
insert into sigin values(1,'2017-01-01');
insert into sigin values(1,'2017-01-02');
insert into sigin values(1,'2017-01-03');
insert into sigin values(1,'2017-01-04');
insert into sigin values(2,'2017-01-01');
insert into sigin values(2,'2017-01-02');
insert into sigin values(2,'2017-01-03');
insert into sigin values(1,'2017-01-10');
insert into sigin values(2,'2017-01-10');
insert into sigin values(1,'2017-01-11');
insert into sigin values(2,'2017-01-11');
insert into sigin values(1,'2017-01-12');
insert into sigin values(2,'2017-01-12');
commit;
大体思路如下:
- 首先根据userid进行分组,将用户所有活跃的记录依次按照时间排序并标上序号。
- 根据时间有序的特点,将所有时间减去它对应的序号,获取连续活跃时间唯一的时间点。如
2017-01-01 1
2017-01-02 2
2017-01-03 3
时间减去序号,得唯一时间2016-12-31。
- 根据userid和这个连续活跃时间唯一的时间点进行分组,计算连续活跃天数。
-- 每个用户的几段连续活跃的天数
select
userid,
to_date(sigindate,'yyyy-mm-dd')-sigin_rank as date_rank,
count(1) as sigincount
from
(
select
userid,
sigindate,
row_number() over(partition by userid order by sigindate) as sigin_rank
from sigin
) c group by userid, to_date(sigindate,'yyyy-mm-dd')-sigin_rank;
得到结果1如下,
USERID DATE_RANK SIGINCOUNT
1 2017/1/5 3
2 2017/1/6 3
1 2016/12/31 4
2 2016/12/31 3
上述方法可以找到每个用户的连续活跃天数,但用户中间有中断时程序就无法满足,一个用户出现了多条记录,分别为用户的多段连续活跃所产生。
我们最终的目标是得到用户连续活跃的最大天数,可利用上述方法所得到的结果,在外面再嵌套一层,针对userid进行group by,得到每个用户的最大活跃天数。
select d.userid, Max(d.sigincount) as max_sigincount from (
select
userid,
to_date(sigindate,'yyyy-mm-dd')-sigin_rank as date_rank,
count(1) as sigincount
from
(
select
userid,sigindate,row_number() over(partition by userid order by sigindate) as sigin_rank
from sigin
) c group by userid ,to_date(sigindate,'yyyy-mm-dd')-sigin_rank
) d group by d.userid
得到结果2如下,
USERID MAX_SIGINCOUNT
1 4
2 3
如果还需要得到用户连续活跃最大天数中这一段的首次活跃时间,可以把以上两个结果进行关联得到。
-- 每个用户连续活跃的最大天数和连续活跃的第一天的时间
select f.userid,g.date_rank+1,f.max_sigincount from (
select d.userid, Max(d.sigincount) as max_sigincount from (
select
userid,
to_date(sigindate,'yyyy-mm-dd')-sigin_rank as date_rank,
count(1) as sigincount
from
(
select
userid,sigindate,row_number() over(partition by userid order by sigindate) as sigin_rank
from sigin
) c group by userid ,to_date(sigindate,'yyyy-mm-dd')-sigin_rank
) d group by d.userid
) f inner join (
select
userid,
to_date(sigindate,'yyyy-mm-dd')-sigin_rank as date_rank,
count(1) as sigincount
from
(
select
userid,sigindate,row_number() over(partition by userid order by sigindate) as sigin_rank
from sigin
) c group by userid ,to_date(sigindate,'yyyy-mm-dd')-sigin_rank
) g on f.userid = g.userid and f.max_sigincount = g.sigincount;
得到结果3如下,
USERID G.DATE_RANK+1 MAX_SIGINCOUNT
2 2017/1/7 3
1 2017/1/1 4
2 2017/1/1 3
结果3还存在一个问题,如果用户有两段连续活跃的天数相同且最大,则第二段连续活跃的首次活跃时间是不对的,这个问题怎么解决呢?欢迎留言你的解决方案。
本文首发于steem,感谢阅读,转载请注明。
微信公众号「padluo」,分享数据科学家的自我修养,既然遇见,不如一起成长。
读者交流电报群
知识星球交流群
Congratulations @padluo! You received a personal award!
You can view your badges on your Steem Board and compare to others on the Steem Ranking
Do not miss the last post from @steemitboard:
Vote for @Steemitboard as a witness to get one more award and increased upvotes!
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit
Congratulations @padluo! You received a personal award!
Click here to view your Board
Downvoting a post can decrease pending rewards and make it less visible. Common reasons:
Submit