SQL进阶day10————多表查询

目录

1嵌套子查询

1.1月均完成试卷数不小于3的用户爱作答的类别

1.2月均完成试卷数不小于3的用户爱作答的类别

​编辑1.3 作答试卷得分大于过80的人的用户等级分布

2合并查询

2.1每个题目和每份试卷被作答的人数和次数

2.2分别满足两个活动的人

3连接查询

3.1满足条件的用户的试卷完成数和题目练习数

3.2 每个6/7级用户活跃情况

1嵌套子查询

1.1月均完成试卷数不小于3的用户爱作答的类别

我的代码：思路就是这么个思路，反正没有搞出来当月均完成试卷数

select tag,count(submit_time) tag_cnt
from exam_record er join examination_info ei
on er.exam_id = ei.exam_id
where uid in (当月均完成试卷数>=3)
group by tag
order by tag_cnt desc

反正没有搞出来当月均完成试卷数，报错： 

大佬正确答案：

居然和我的差不多，我就分组的时候少了uid，还有按照uid进行分组。此外，作答次数=count(start_time)，而不是提交次数。

select tag, count(start_time) as tag_cnt
from exam_record er inner join examination_info ei
on er.exam_id = ei.exam_id
where uid in 
(select uid
from exam_record er 
group by uid, month(start_time)
having count(submit_time) >= 3)
group by tag
order by tag_cnt desc

复盘：

（1）uid,month(submit_time)是啥呢，如果原来只是按照month(submit_time)进行分组，1002,1003,1005都有多个

 （2）如果按照uid,month(submit_time)进行分组，情况如下

（3）这么如果只是按照month(submit_time) 分组，uid,month(submit_time)只有9和null两种情况，当使用GROUP BY子句时，NULL值将被视为一个独立的分组，并在结果集中显示一个额外的分组来表示它。

（4）结果显示只有1002,1005这两个用户满足要求，然后查找这两个用户的作答的类别及作答次数。

（5）验证：where uid =1002 or uid = 1005  等价于 子查询的效果

还有一种大佬做法是：

select tag,count(start_time) tag_cnt
from exam_record er join examination_info ei
on er.exam_id = ei.exam_id
-- where uid =1002 or uid = 1005 
WHERE er.uid IN (SELECT uidFROM exam_recordGROUP BY uidHAVING COUNT(submit_time) / COUNT(DISTINCT DATE_FORMAT(submit_time, "%Y%m")) >= 3
)
group by tag
order by tag_cnt desc

 这样出来的两个用户也是1002和1005：

• 相当于：月均完成试卷数 = 总完成次数/哪些月份提交了数据

COUNT(DISTINCT DATE_FORMAT(submit_time, "%Y%m"))=1，所以答案一样的。

COUNT(DISTINCT DATE_FORMAT(submit_time, "%Y%m"))中的distinct很重要：

1.2月均完成试卷数不小于3的用户爱作答的类别

我的代码：答案错误，但是我能发现的的改了，

（1）SQL类，（2）当天，（3）作答人数

select er.exam_id,
any_value(count(er.submit_time)) uv,
round(avg(er.score),1) avg_score
from examination_info ei join exam_record er
on ei.exam_id = er.exam_id
where ei.tag = "SQL" 
and day(submit_time)=day(release_time)
and er.uid in 
(select uid 
from user_info
where level > 5)
group by er.exam_id
order by uv desc,avg_score asc

正确代码：

select er.exam_id,
any_value(count(distinct er.uid)) uv,
round(avg(er.score),1) avg_score
from examination_info ei join exam_record er
on ei.exam_id = er.exam_id
where ei.tag = "SQL" 
and date_format(submit_time,'%Y%m%d')=date_format(release_time,'%Y%m%d')
and er.uid in 
(select uid 
from user_info
where level > 5)
group by er.exam_id
order by uv desc,avg_score asc

复盘：

（1）同一天，不能用day函数，0901和0201的day都是1，但是不是同一天。

（2）计算人数时，要加distinct才对：

原数据有这种离谱的情况？？

1.3 作答试卷得分大于过80的人的用户等级分布

我的正确代码：直接三表连接

select level,count(level) level_cnt
from user_info u 
join exam_record er
on u.uid = er.uid
join examination_info ei
on ei.exam_id = er.exam_id
where ei.tag = 'SQL'
and er.score>80
group by level

嵌套子查询的方法代码：

SELECT level, 
COUNT(level) AS level_cnt
FROM user_info
WHERE uid IN (SELECT DISTINCT uidFROM exam_recordWHERE score > 80AND exam_id IN (SELECT exam_id FROM examination_info WHERE tag = 'SQL'))
GROUP BY level
ORDER BY level_cnt DESC;

2合并查询

2.1每个题目和每份试卷被作答的人数和次数

我的代码：分别查询然后用union all合并起来，但是答案错了

select exam_id tid,
count(distinct er.uid) uv,
count(distinct pr.submit_time) pv
from exam_record er join practice_record pr
using(uid)
group by exam_idunion allselect question_id tid,
count(distinct er.uid) uv,
count(distinct pr.submit_time) pv
from exam_record er join practice_record pr
using(uid)
group by question_id

正确答案：

select * from 
(SELECT exam_id tid,count(DISTINCT uid) uv,count(uid) pv from exam_record
group by exam_id
order by uv desc,pv desc)a
UNION ALL
SELECT * FROM
(SELECT question_id tid,count(DISTINCT uid) uv,count(uid) pv from practice_record
GROUP BY question_id
order by uv desc,pv desc)b

我的代码改正：这个题最后不要合并，题目和试卷在不同的表里，分别查询在合并就好了

select exam_id tid,
count(distinct er.uid) uv,
count(er.uid) pv
from exam_record er
group by exam_idunion allselect question_id tid,
count(distinct pr.uid) uv,
count(pr.uid) pv
from practice_record pr
group by question_id

还没排序：

但是使用 union 和 多个order by 不加括号 【报错】，order by 在 union 连接的子句不起作用，但是在子句的子句中起作用。

方法一：所以加两个order的话正确要这样写：

#正确代码
select * from 
(
select exam_id as tid,count(distinct uid) as uv,count(uid) as pv
from exam_record a
group by exam_id
order by uv desc, pv desc
) a
union
select * from 
(
select question_id as tid,count(distinct uid) as uv,count(uid) as pv
from practice_record b
group by question_id
order by uv desc, pv desc
) attr

方法二：或者利用left(str,length) 函数： str左边开始的长度为 length 的子字符串，在本例中为‘9’和‘8’。

order by left(tid,1) desc,uv desc,pv desc

解释：试卷编号以‘9’开头、题目编号以‘8’开头，对编号进行降序就是对"试卷"和"题目"分别进行排序。

(#每份试卷被作答的人数和次数selectexam_id as tid,count(distinct uid) as uv,count(*) as pv
from exam_record
group by exam_id
)
union
(#每个题目被作答的人数和次数selectquestion_id as tid,count(distinct uid) as uv,count(*) as pv
from practice_record
group by question_id
)
#分别按照"试卷"和"题目"的uv & pv降序显示
order by left(tid,1) desc,uv desc,pv desc

2.2分别满足两个活动的人

我的垃圾代码：不知道新的值怎么弄

(select uidfrom exam_recordgroup by 1001having score>85
)tselect uid	t.activity
from examination_info ei join exam_record er
on ei.exam_id = er.exam_id

大佬代码：

(select uid,'activity1' as activity
from exam_record er
where year(start_time)='2021'
group by uid
having min(score)>=85)
union ALL
(select uid,'activity2' as activity
from exam_record er left join examination_info ei on er.exam_id=ei.exam_id
where year(start_time)='2021' and ei.difficulty='hard' and score>=80 
and timestampdiff(second,er.start_time,er.submit_time)<= ei.duration*30
group by uid)
order by uid;

复盘：

（1）select uid,'activity1' as activity...，这样就把activity这一列就设置出来了。

（2）时间差函数：timestampdiff，如计算差多少分钟，timestampdiff(minute,时间1,时间2)，是时间2-时间1，单位是minute。

这里是至少有一次用了一半时间就完成：

完成时间<=考试时长/2 （单位为分钟minute）

完成时间<=考试时长*60/2 =考试时长*30（单位为秒second）

timestampdiff(second,er.start_time,er.submit_time)<= ei.duration*30

（3）每次试卷得分都能到85分，相当于最低分min>=85

3连接查询

3.1满足条件的用户的试卷完成数和题目练习数

我的报错代码：看来不是这么简单粗暴的事情 

select u.uid,
count(er.submit_time) exam_cnt,
count(pr.submit_time) question_cnt
from user_info u join exam_record er
on u.uid = er.uid
join practice_record pr
on pr.uid = u.uid
join examination_info ei
on ei.exam_id = er.exam_id
where year(er.submit_time)='2021'
group by u.uid
having ei.tag = 'SQL'
and ei.difficulty = 'hard'
and u.level = 7
and avg(er.score)>80

正确代码：

# select er.uid as uid,
# count(distinct er.submit_time) as exam_cnt,
# count(distinct pr.submit_time) as question_cnt
select er.uid as uid,
count(distinct er.exam_id) as exam_cnt,
count(distinct pr.id) as question_cntfrom exam_record er 
left join practice_record pr 
on er.uid=pr.uid 
and year(er.submit_time)=2021 
and year(pr.submit_time)=2021where er.uid in(select er.uidfrom exam_record er left join examination_info ei on er.exam_id = ei.exam_idleft join user_info ui on er.uid = ui.uid where tag='SQL' and difficulty='hard' and level = 7group by er.uidhaving avg(score) > 80) 
group by er.uid
order by exam_cnt,question_cnt desc

复盘：

有4个表，很多个条件

（1）先通过子查询中连接，er,ui和ei筛选出高难度SQL试卷得分平均值大于80并且是7级的红名大佬（返回用户uid）

（2） 再统计这些大佬的2021年试卷总完成次数，和题目总练习次数

（3）注意第（2）步中连接是左连接，不应该出现试卷为null，题目不为null的情况！

from exam_record er left join practice_record pr

（4）不懂为什么不能用 er.submit_time， pr.submit_time来计算

# select er.uid as uid,
# count(distinct er.submit_time) as exam_cnt,
# count(distinct pr.submit_time) as question_cnt
select er.uid as uid,
count(distinct er.exam_id) as exam_cnt,
count(distinct pr.id) as question_cnt

3.2 每个6/7级用户活跃情况

我的错误代码：

总活跃月份数？其他都是2021年的，活跃是啥意思？

select er.uid,
# act_month_total,
count(er.start_time) act_days_2021,
count(er.submit_time) act_days_2021_exam,
count(pr.submit_time) act_days_2021_question
from exam_record er left join practice_record pr
on er.uid=pr.uid
where year(er.submit_time)=2021
and er.uid in
(select uid 
from user_info
where level = 7 or level = 6)
group by er.uid

正确代码

selectuser_info.uid,count(distinct act_month) as act_month_total,count(distinct casewhen year (act_time) = '2021' then act_dayend) as act_days_2021,count(distinct casewhen year (act_time) = '2021'and tag = 'exam' then act_dayend) as act_days_2021_exam,count(distinct casewhen year (act_time) = '2021'and tag = 'question' then act_dayend) as act_days_2021_question
from(SELECTuid,exam_id as ans_id,start_time as act_time,date_format (start_time, '%Y%m') as act_month,date_format (start_time, '%Y%m%d') as act_day,'exam' as tagfromexam_recordUNION ALLselectuid,question_id as ans_id,submit_time as act_time,date_format (submit_time, '%Y%m') as act_month,date_format (submit_time, '%Y%m%d') as act_day,'question' as tagfrompractice_record) totalright join user_info on total.uid = user_info.uid
whereuser_info.level in (6, 7)
group byuser_info.uid
order byact_month_total desc,act_days_2021 desc

复盘

（1）case when是关键

（2）2021年活跃天数 = 2021年试卷作答活跃天数 + 2021年答题活跃天数

则 exam as tag 和 practice as tag，自定义一列，为了区分是考试还是练习，便于区别计算

（3）右连接 total     right join user_info      on total.uid = user_info.uid

因为自组合的total表：没有1003

原本的user_info表：

但是6/7级的大佬中是有1003的