常用SQL案例

Posted 2023-03-02 不务正业RD从产品经理转行到数据分析

一、开窗函数

1.开窗函数求占比

select
    city_name,
    type,
    id_cnt_by_city_name,
    count(distinct id) id_cnt, --type分组ID数
    
    count(distinct id) / id_cnt_by_b as rate --占比
from
    (
        select
            type,
            city_name,
            id,
            count(id) over (partition by city_name) id_cnt_by_city_name --求城市总id数
        from
            table_Name
    ) xxx
group by
    city_name,
    type,
    id_cnt_by_city_name

2.随机抽样 

-- 根据type_1,每组抽样600 
select 
    *
from 
(
    select
        id,
        type_1,
        type_2,
        row_number() over(partition by type_1 order by rand()) as rn   --rand() 获取随机数， order by排序，row_number() 排名次序  
    from
        table_name
)a 
where rn <= 600 --抽样600  

二、josn数据解析

       学会Hive解析Json数组_000X000的博客-CSDN博客_hive解析json