Teradata按某字段分组,排序后,取每组第一条记录

最新推荐文章于 2023-07-21 14:45:48 发布

celebrateyang

最新推荐文章于 2023-07-21 14:45:48 发布

阅读量2.1k

点赞数

文章标签： Teradata

本文链接：https://blog.csdn.net/celebrateyang/article/details/93739010

版权

创建临时测试表脚本如下:

CREATE MULTISET VOLATILE TABLE temprecommend 
     (
      item_id DECIMAL(18,0),
      rankScore DECIMAL(3,2),
      sourcingTaskId BYTEINT
	  )
PRIMARY INDEX ( item_id )
ON COMMIT PRESERVE ROWS;

初始化数据脚本如下:

insert into temprecommend (item_id,rankScore,sourcingTaskId) values(152241683582,7.50,116);
insert into temprecommend (item_id,rankScore,sourcingTaskId) values(152241683581,7.10,116);
insert into temprecommend (item_id,rankScore,sourcingTaskId) values(152055889769,7.30,115);
insert into temprecommend (item_id,rankScore,sourcingTaskId) values(152004277613,7.20,115);
insert into temprecommend (item_id,rankScore,sourcingTaskId) values(152241683585,7.40,116);

SQL 如下:

select * from 
(select row_number() over (partition by sourcingtaskid order by rankscore desc) as num,item_id,rankScore,sourcingTaskId
 from temprecommend) as data
where data.num = 1

解释:1.下面一句实现按sourcingtaskid分组,并按rankscore 排序

select row_number() over (partition by sourcingtaskid order by rankscore desc) as num,item_id,rankScore,
sourcingTaskId from temprecommend

效果下图:
在这里插入图片描述
2. 下面一句实现从所有分组中只取第一条记录

select * from 
(select row_number() over (partition by sourcingtaskid order by rankscore desc) as num,item_id,rankScore,sourcingTaskId
 from temprecommend) as data
where data.num = 1