chenzhh25的博客

1、行转列数据源：要实现的效果：Hive：collect_set转为数组并去重，concat_ws将数组用逗号间隔连接成字符串select user_id , concat_ws(',', collect_set(order_id)) as order_idsfrom tmp.tmp_row_to_colwhere 1 = 1group by user_id...

select greatest(1, 2, 3, 4) ; 结果：4select least(1, 2, 3, 4) ; 结果：1select greatest(1, 2, null, 3, 4) ; 结果：空select least(1, 2, null, 3, 4) ; 结果：空Hive和Presto通用，值得注意的是，用这两个函数时，各列不能存在null值，也最好都...

row_numberrow_number() over(partititon by col1 order by col2) as rank_num1结果：1,2,3,4,5rankrank() over(partititon by col1 order by col2) as rank_num2结果：1,2,2,4,5dense_rankdense_rank() ove...

-- 1、方式一select country_nmfrom dw_pub_country_tdwhere country_nm = 'Lao People''s Democratic Republic'-- 2、方式二select country_nmfrom dw_pub_country_tdwhere country_nm = "Lao People's Democratic...

用个排序的窗口函数即可实现：select distinct user_idfrom ( select user_id , continue_date , count(login_date) as continue_day_cnt from ( select user_id , login_date ...