SQL统计与分析技巧-优快云博客

本文链接：https://blog.youkuaiyun.com/Sun123234/article/details/126545753

题型1：最大同时在线人数

SQL163
在这里插入图片描述
思路：
1，进入时刻记为1，出来时间记为-1.连接起来
2，SUM窗口函数，按文章id维度，统计按时间戳升序的观看人数变化情况
3，最外层SELECT按artical_id聚合，通过MAX（cnt）取出瞬时观看最大值max_uv，并排序。
同时在线人数的题型都是这个步骤！SQL189同题型

题型2：union的使用

不是连接，是直接上下摞起来，用union
SQL 135
在这里插入图片描述

select u_i.uid as uid,
       count(distinct act_month) as act_month_total,
       count(distinct case 
             when year(act_time) = 2021 
             then act_day 
             end) as act_days_2021,
       count(distinct case 
             when year(act_time) = 2021 
             and tag = 'exam' 
             then act_day 
             end) as act_days_2021_exam,
        count(distinct case
             when year(act_time) = 2021
             and tag = 'question'
             then act_day
             end) as act_days_2021_question
from user_info u_i
left join (select uid,
             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 tag
      from exam_record
      union all 
      select uid,
             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 tag
      from  practice_record
      ) exam_and_practice
on exam_and_practice.uid = u_i.uid
where u_i.level >= 6
group by uid
order by act_month_total desc, act_days_2021 desc

1，试卷数据与题目数据用union all 连接起来，且用tag标注是来自哪张表
2，count(distinct case when 条件 then 取的结论数据列 end)

题型3：如何找到每个月的新增用户

给原表加一个标记新用户的tag列，为1 则是新用户，0为旧用户
IF(start_time=MIN(start_time)OVER(PARTITION BY uid),1,0) AS tag
这样sum(tag) 能直接计算得新增用户数
例题：
SQL144月活用户数、新增用户数、截止当月的单月最大新增用户数、截止当月的累积用户数
在这里插入图片描述

select *,max(month_add_uv)over( order by ym range between unbounded preceding and current row),
sum(month_add_uv)over( order by ym range between unbounded preceding and current row)
from
(select ym,count(distinct uid) as mau,sum(tag) as month_add_uv
from
(select *,date_format(start_time,'%Y%m')as ym,
if(start_time=min(start_time)over(partition by uid),1,0) as tag
from exam_record ) a
group by ym ) b

月活用户数 — count(distinct uid)
新增用户数 —sum(tag)新增列求和
截止当月的单月最大新增用户数 — max(新增用户数)窗口函数range between unbounded preceding and current row
截止当月的累积用户数 — sum(新增用户数)窗口函数range between unbounded preceding and current row

4，Exists的使用。如果满足条件输出数据1，如果不满足条件则输出另一部分数据2

SQL149
在这里插入图片描述

with t as (
select uid,ui.level,
    count(start_time) - count(submit_time) as incomplete_cnt,
    round(ifnull(1-count(submit_time)/count(start_time),0),3) as incomplete_rate,
    count(start_time) as total_cnt
from user_info ui left join exam_record using(uid)
group by uid)

select uid,incomplete_cnt,incomplete_rate
from t 
where exists(select uid from t where t.level = 0 and incomplete_cnt>2)
and level =0
union all
select uid,incomplete_cnt,incomplete_rate
from t 
where not exists(select uid from t where t.level = 0 and incomplete_cnt>2)
and total_cnt > 0
order by incomplete_rate;

with t as () —把要的数据字段列表示出来备用
select 数据1 where exists(条件)
union all
select 数据2 where not exists(条件)

5，取排前几次的数据的方法：

1）排序后limit n
2）row_number()over(）窗口函数排序成cn，再取cn <= n

6,select from (select **** ) a

from (新建表中) a 要另取名字啊

7，日期函数

datediff(时间1，时间2)：计算两个日期之间间隔的天数，单位为日
timestampdiff(时间单位，开始时间，结束时间):两个日期的时间差，返回的时间差形式由时间单位决定(日，周，月，年)
date_add(日期，interval n 时间单位) ：返回加上n个时间单位后的日期
date_sub(日期，interval n 时间单位 )：返回减去n个时间单位后的日期
date_format(时间，‘%Y-%m-%d’)：强制转换时间为所需要的格式

8，SQL138连续两次作答试卷的最大时间窗

在这里插入图片描述

with t2 as (
select uid,count(start_time) total,  -- 用户2021年作答的次数
datediff(max(start_time),min(start_time))+1 diff_time, -- 头尾作答时间窗
max(datediff(next_time,start_time))+1 days_window  -- 最大间隔天数
from (select uid,start_time,lead(start_time,1)over(partition by uid order by start_time) as next_time
      from exam_record
      where year(start_time) = 2021
     ) t1
group by uid
)

select uid,days_window,round(total*days_window/diff_time,2) avg_exam_cnt
from t2
where diff_time >1
order by days_window desc,avg_exam_cnt desc ;

1）要求同个用户的前后时间间隔，用到lead(start_time,1)over(partition by uid order by start_time) as next_time ，然后datediff（）做差
2）用with t as ()做备用，避免多次嵌套

9，排序函数

rank() over() 1 2 2 4 4 6 (计数排名，跳过相同的几个，eg.没有3没有5)
row_number() over() 1 2 3 4 5 6 (赋予唯一排名)
dense_rank() over() 1 2 2 3 3 4 (不跳过排名，可以理解为对类别进行计数)
percent_rank() over() 按照数字所在的位置进行百分位分段
ntile(n)over() 将数字按照大小平均分成n段
lead(字段名，n)over()把字段数据向前移n个单元格
lag(字段名，n)over()把字段数据向后移n个单元格
SQL140 、SQL141、SQL164（计算次日留存率）