在Oracle中按组不使用聚合函数或行到列

Posted 2023-05-09

【中文标题】在Oracle中按组不使用聚合函数或行到列

【英文标题】：Pivot without aggregate function or Row to Columns by group in Oracle

【发布时间】：2016-12-21 09:36:39

【问题描述】：

我正在尝试按行类型将行转换为列。 这里给出 Table_1 Table_1

           Table_1
CITY            AMOUNT      TYPE_ID
Moscow         158000          1
New York       94500           1
Moscow         1478000         2
Los Angeles    162000          2
New York       5500000         2
Los Angeles    35400           1
Moscow         741200          1

并在结果中使用选择脚本，我想在 Table_2 Table_2 中采用类似的方法

            Table_2
CITY           TYPE_1_AMOUNT       TYPE_2_AMOUNT
Moscow           158000               1478000
Moscow           741200                  NULL
New York         94500                5500000
Los Angeles      35400                162000

我尝试使用 PIVOT。但必须有聚合函数。 聚合函数 MAX() 只检索最大数量...

【问题讨论】：

您好，请将图片替换为文字。把它用逗号/制表符作为分隔符，我稍后会格式化它

【参考方案1】：

select      city
           ,min (case type_id when 1 then amount end)   as type_1_amount
           ,min (case type_id when 2 then amount end)   as type_2_amount

from       (select      city,type_id,amount

                       ,row_number () over 
                        (
                            partition by    city,type_id
                            order by        amount
                        ) as rn

            from        Table_1
            )

group by    city
           ,rn

order by    city
           ,rn               

---

+-------------+---------------+---------------+
| CITY        | TYPE_1_AMOUNT | TYPE_2_AMOUNT |
+-------------+---------------+---------------+
| Los Angeles | 35400         | 162000        |
+-------------+---------------+---------------+
| Moscow      | 158000        | 1478000       |
+-------------+---------------+---------------+
| Moscow      | 741200        | (null)        |
+-------------+---------------+---------------+
| New York    | 94500         | 5500000       |
+-------------+---------------+---------------+

【参考方案2】：

在 Oracle 11.1 及更高版本中，您可以使用 PIVOT 运算符执行相同操作 - 但首先您必须使用 row_number() 之类的东西来区分行（与 Dudu 的解决方案相同）。 PIVOT 解决方案如下所示：

with
     table_1 ( city, amount, type_id ) as (
       select 'Moscow'     ,  158000, 1 from dual union all
       select 'New York'   ,   94500, 1 from dual union all
       select 'Moscow'     , 1478000, 2 from dual union all
       select 'Los Angeles',  162000, 2 from dual union all
       select 'New York'   , 5500000, 2 from dual union all
       select 'Los Angeles',   35400, 1 from dual union all
       select 'Moscow'     ,  741200, 1 from dual
     )
-- end of test data; SQL query begins below this line
select city, type_1, type_2
from   ( select city, amount, type_id,
                row_number() over (partition by city, type_id order by amount) as rn
         from   table_1
       )
pivot  ( min(amount) for type_id in (1 as type_1, 2 as type_2) )
order by city, type_1, type_2    --  ORDER BY is optional
;

CITY            TYPE_1     TYPE_2
----------- ---------- ----------
Los Angeles      35400     162000
Moscow          158000    1478000
Moscow          741200
New York         94500    5500000

4 rows selected.