SQL26 每个6/7级用户的活跃情况(表的连接方式的选取)
每个6/7级用户活跃情况
https://www.nowcoder.com/practice/a32c7c8590324c96950417c57fa6ecd1?tpId=240&tags=&title=&difficulty=0&judgeStatus=0&rp=0&sourceUrl=%2Fexam%2Foj%3Fpage%3D1%26tab%3DSQL%25E7%25AF%2587%26topicId%3D240
我的稀碎答案
(错因分析及待改进的地方)
select uid,count(distinct month0) as act_month_total,
count(distinct if(substring(month0,1,4)=2021,date0,null)) as act_days_2021,
%%%应用substring筛选太复杂
count(distinct if(substring(month0,1,4)=2021 and
substring(tid,1)=9,date(er.submit_time,null)))
as act_days_2021_exam,
%%%应用substring筛选太复杂
count(distinct if(substring(month0,1,4)=2021 and
substring(tid,1)=8,date(pr.submit_time,null)))
as act_days_2021_question
%%%应用substring筛选太复杂
from
(select uid,exam_id as tid
date_format(submit_time,'%Y%m') as month0,date(submit_time) as date0,
%%%题意理解有误,作答不提交也算活跃
from exam_record as er
where score is not null
union all
select uid,question_id as tid,
date_format(submit_time,'%Y%m') as month0,
date(submit_time) as date0
from practice_record as pr
)
where uid in %%%这会使得最终表中只存在有作答记录的6,7级用户
(select uid
from user_info
where level=6 or level=7)
group by tid
order by act_month_total desc,act_days_2021 desc
分析:
(此部分的代码块部分仅涉及关键语句,不可直接应用)
筛选对象——6/7级用户
- 筛选方式:left join 而非where uid in,以防筛去无作答记录的6/7级用户
- 筛选语句:left join…on ui.uid=q.uid where level=6 or level=7 (不知道为什么在on里筛选等级不奏效)
筛选变量——总活跃月份数、2021年活跃天数、2021年试卷作答活跃天数、2021年答题活跃天数
需要注意的点:
(1) 仅后三个变量有2021的限制,所以2021不能在提取数据时就进行筛选,应在变量生成时筛选;
(2) 对于2021年活跃天数,由于存在同一天同时存 在试卷和习题作答记录,要进行合并与去重,相应地,计算后两个变量时要予以区分
- 总活跃月份数——
date_format(submit_time,'%Y%m') as month0
+date_format(submit_time,'%Y%m') as month0
→count(distinct month0) as act_month_total
- 2021年活跃天数——
start_time as act_time
+submit_time as act_time
count(distinct if(year(act_time)=2021,date0,null)) as act_days_2021
- 2021年试卷作答活跃天数——
'exam' as tag
+'question' as tag
count(distinct if(year(act_time)=2021 and tag='exam',date0,null))
as act_days_2021_exam
- 2021年答题活跃天数——
count(distinct if(year(act_time)=2021 and tag='question',date0,null))
as act_days_2021_question
排序方式——按照总活跃月份数、2021年活跃天数降序排序
order by act_month_total desc,act_days_2021 desc
提交答案
select ui.uid,count(distinct month0) as act_month_total,
count(distinct if(year(act_time)=2021,date0,null)) as act_days_2021,
count(distinct if(year(act_time)=2021 and tag='exam',date0,null))
as act_days_2021_exam,
count(distinct if(year(act_time)=2021 and tag='question',date0,null))
as act_days_2021_question
%%%在下表中增加了tag和act_time两个变量,简化了2021年的筛选,对exam和question记录也做了区分
from user_info as ui
left join %%%保证6、7级以上暂无活跃记录的用户也在表中
(select uid,'exam' as tag,start_time as act_time,
date_format(start_time,'%Y%m') as month0,date(start_time) as date0
%%%作答不提交也算活跃
from exam_record as er
union all
select uid,'question' as tag,submit_time as act_time,
date_format(submit_time,'%Y%m') as month0,
date(submit_time) as date0
from practice_record as pr
) as q
on ui.uid=q.uid
where level=6 or level=7 %%%筛选6,7级用户
group by uid
order by act_month_total desc,act_days_2021 desc