FIRST_VALUE没有像我期望的那样使用PARTITION BY

Here is my test set:

这是我的测试集：

master_ref  ref          value
56279276    56279325    FRAME ASSEMBLY1
56279276    384062724   FRAME ASSEMBLY2
56279276    443450775   FRAME ASSEMBLY3

I want to retrieve the value field based on the highest ref given a master_ref.

我想根据给定master_ref的最高引用来检索值字段。

Here is what I tried that just return everything:

这是我尝试过的只返回一切：

select first_value(value) over (partition by ida3masterreference order by ida3a4 desc) value, ida3masterreference, ida3a4 from sts_epm_title1;

I expected to only get one result:

我希望只得到一个结果：

master_ref    ref         value
56279276    443450775   FRAME ASSEMBLY3

But it still returns all 3 results. What am I doing wrong? Thanks!

但它仍然返回所有3个结果。我究竟做错了什么？谢谢！

2 个解决方案

#1

first_value() is an analytic function, so it does not reduce the number of rows. You apparently want an aggregation function so use the keep syntax:

first_value（）是一个分析函数，因此它不会减少行数。你显然想要一个聚合函数，所以使用keep语法：

select max(value) keep (dense_rank first order by ida3a4 desc) as value, 
       ida3masterreference, max(ida3a4) as ida3a4
from sts_epm_title1
group by ida3masterreference

#2

You could use ROW_NUMBER/RANK:

你可以使用ROW_NUMBER / RANK：

WITH cte AS (
   select row_number() over (partition by ida3masterreference 
                             order by ida3a4 desc) AS rn, t.*
   from sts_epm_title1 t
)
SELECT *
FROM cte
WHERE rn = 1;

#1