题解 | #统计活跃间隔对用户分级结果#
统计活跃间隔对用户分级结果
https://www.nowcoder.com/practice/6765b4a4f260455bae513a60b6eed0af
1.需要首先整理出每个用户uid的最早活跃时间和最晚活跃时间:min(out_time)和max(out_time),group by uid (命名为表1)
2.今天:select max(date(out_time)) from 原表 ,也就是整个表数据中的最大日期
3.划分等级,当用户最晚活跃时间距离今天>=30天,则为流失用户,>=7天,则为沉睡用户,<7天并且最早活跃时间距离今天<7天,为新增用户,否则为忠实用户
4.计算比例,按照等级分组,count(uid)即为每个等级的用户数,select count(distinct uid from 原表)即为总的用户数,相除为每个等级的用户占比
5.细节:保留小数位和降序排序
select case when datediff((select date(max(out_time))from tb_user_log),uv_date)>=30 then '流失用户' when datediff((select date(max(out_time))from tb_user_log),uv_date)>=7 then '沉睡用户' when datediff((select date(max(out_time))from tb_user_log),uv_date)<7 and datediff((select date(max(out_time))from tb_user_log),new_dt)>=7 then '忠实用户' else '新晋用户' end as user_grade,round(count(uid)/(select count(distinct uid)from tb_user_log),2) as ratio from( select uid,max(date(out_time)) as uv_date,min(date(out_time)) as new_dt from tb_user_log group by uid) t group by user_grade order by ratio desc