当值与其他GROUP BY子句不同时，Oracle SQL选择SUM

我有以下示例_数据集：

ORD |   POS   | INSGRP | INS |   DELI    |   DELPOS
-------------------------------------------------------
1   | 'Pos11' | '11'   | 11  | 'deli111' | 'DelPos1111'
1   | 'Pos11' | '11'   | 11  | 'deli112' | 'DelPos1121'
1   | 'Pos11' | '11'   | 11  | 'deli112' | 'DelPos1122'

2   | 'Pos21' | '21'   | 21  | 'deli211' | 'DelPos2111'
2   | 'Pos21' | '22'   | 22  | 'deli221' | 'DelPos2211'
2   | 'Pos21' | '22'   | 22  | 'deli221' | 'DelPos2212'

3   | 'Pos31' | '31'   | 31  | 'deli311' | 'DelPos3111'
3   | 'Pos32' | '31'   | 31  | 'deli321' | 'DelPos3211'
3   | 'Pos32' | '31'   | 31  | 'deli321' | 'DelPos3212'
3   | 'Pos32' | '31'   | 31  | 'deli322' | 'DelPos3221'

• 当ORD有多个POS时，只有一个INSGRP
• 当ORD只有一个POS时，可以有多个INSGRP
• 这是我实际请求的一个简单版本，只暴露了这个问题
• 一个餐厅可以有很多熟食店
• 一个熟食店可以有很多熟食店
• DELI和DELPOS用于添加重复ORD-POS-INSGRP-INS的行

我想根据ORD、POS和INSGRP对INS列求和

• 当POS只有一个INSGRP时，我希望INS保持不变
• 当一个POS有多个INSGRP时，我需要该POS的不同INSGRP的INS之和

因此，我希望得到以下结果：

ORD | SUM(INS)
-----------------------------------------------------------------------
1   | 11
One INSGRP for one POS, so INS stay the same 11
2   | 43
Two INSGRP for one POS, so INS is the SUM of the INSGRP 21 and 22 so 43 
3   | 31
One INSGRP for two POS, so INS stay the same 31

我尝试了以下方法：

SELECT ord,
       SUM (ins) AS ins
FROM sample_data
GROUP BY ord, pos
ORDER BY ord;

得到了这个结果：

ORD | INS
---------
1   | 33
2   | 65
3   | 31
3   | 93

所有的东西都是总和，ORD 3的每个INSGRP都有一行。然后我将INS添加到GROUPBY子句中，但它只在两行上分隔ORD 2

我尝试将所有内容分组，并在开头添加一个独特的：

SELECT UNIQUE ord,
       SUM (ins) AS ins
FROM sample_data
GROUP BY ord, pos, deli, DelPos
ORDER BY ord;

结果是：

ORD | INS
---------
1   | 11
2   | 21
2   | 22
3   | 31

这是我得到的最接近我想要的结果。唯一缺少的部分是将ORD 2分组在一行上，并对INS求和

另外，我希望能够像这样对组进行分区：

SUM (ins) OVER (PARTITION BY ord, pos, deli, DelPos) AS ins

有人能帮我吗

---

以下是样本数据的请求：

WITH sample_data
     AS (SELECT 1 ord, 'Pos11' pos, '11' insGrp, 11 ins, 'deli111' deli, 'DelPos1111' DelPos FROM DUAL
         UNION ALL
         SELECT 1 ord, 'Pos11' pos, '11' insGrp, 11 ins, 'deli112' deli, 'DelPos1121' DelPos FROM DUAL
         UNION ALL
         SELECT 1 ord, 'Pos11' pos, '11' insGrp, 11 ins, 'deli112' deli, 'DelPos1122' DelPos FROM DUAL
         UNION ALL

         SELECT 2 ord, 'Pos21' pos, '21' insGrp, 21 ins, 'deli211' deli, 'DelPos2111' DelPos FROM DUAL
         UNION ALL
         SELECT 2 ord, 'Pos21' pos, '22' insGrp, 22 ins, 'deli221' deli, 'DelPos2211' DelPos FROM DUAL
         UNION ALL
         SELECT 2 ord, 'Pos21' pos, '22' insGrp, 22 ins, 'deli221' deli, 'DelPos2212' DelPos FROM DUAL
         UNION ALL

         SELECT 3 ord, 'Pos31' pos, '31' insGrp, 31 ins, 'deli311' deli, 'DelPos3111' DelPos FROM DUAL
         UNION ALL
         SELECT 3 ord, 'Pos32' pos, '31' insGrp, 31 ins, 'deli321' deli, 'DelPos3211' DelPos FROM DUAL
         UNION ALL
         SELECT 3 ord, 'Pos32' pos, '31' insGrp, 31 ins, 'deli321' deli, 'DelPos3212' DelPos FROM DUAL
         UNION ALL
         SELECT 3 ord, 'Pos32' pos, '31' insGrp, 31 ins, 'deli322' deli, 'DelPos3221' DelPos FROM DUAL)
SELECT UNIQUE ord,
     SUM (ins) AS ins
FROM sample_data
GROUP BY ord, pos, deli, DelPos
ORDER BY ord;

或者创建并插入

create table tbl (ord number, pos varchar2(10),  insGrp varchar2(10), ins number, deli varchar2(10), DelPos varchar2(10))

insert into tbl (SELECT 1 ord, 'Pos11' pos, '11' insGrp, 11 ins, 'deli111' deli, 'DelPos1111' DelPos FROM DUAL
         UNION ALL
         SELECT 1 ord, 'Pos11' pos, '11' insGrp, 11 ins, 'deli112' deli, 'DelPos1121' DelPos FROM DUAL
         UNION ALL
         SELECT 1 ord, 'Pos11' pos, '11' insGrp, 11 ins, 'deli112' deli, 'DelPos1122' DelPos FROM DUAL
         UNION ALL

         SELECT 2 ord, 'Pos21' pos, '21' insGrp, 21 ins, 'deli211' deli, 'DelPos2111' DelPos FROM DUAL
         UNION ALL
         SELECT 2 ord, 'Pos21' pos, '22' insGrp, 22 ins, 'deli221' deli, 'DelPos2211' DelPos FROM DUAL
         UNION ALL
         SELECT 2 ord, 'Pos21' pos, '22' insGrp, 22 ins, 'deli221' deli, 'DelPos2212' DelPos FROM DUAL
         UNION ALL

         SELECT 3 ord, 'Pos31' pos, '31' insGrp, 31 ins, 'deli311' deli, 'DelPos3111' DelPos FROM DUAL
         UNION ALL
         SELECT 3 ord, 'Pos32' pos, '31' insGrp, 31 ins, 'deli321' deli, 'DelPos3211' DelPos FROM DUAL
         UNION ALL
         SELECT 3 ord, 'Pos32' pos, '31' insGrp, 31 ins, 'deli321' deli, 'DelPos3212' DelPos FROM DUAL
         UNION ALL
         SELECT 3 ord, 'Pos32' pos, '31' insGrp, 31 ins, 'deli322' deli, 'DelPos3221' DelPos FROM DUAL)

我想你想要这个：

SELECT ord,
     CASE WHEN COUNT (DISTINCT pos) = 1 THEN SUM(DISTINCT ins) ELSE MIN(ins) END AS ins
FROM sample_data
GROUP BY ord
ORDER BY ord;

ORD INS   
--- ------
  1     11
  2     43
  3     31

---

您能否共享创建脚本并以INSERT语句的形式提供一些示例数据？您所说的“另外，我希望能够分区…”是什么意思？一组人还不够吗？您能给出您的示例数据显示预期的输出吗？正如我所说，这是一个简单的测试，在我的最终请求中，我希望其他字段按不同的列分组。所以我不能按整个选择分组。预期结果是一样的。很好，谢谢！我用分区对它进行了调整，效果很好！当COUNT（DISTINCT pos）OVER（PARTITION BY ord）=1时，SUM（DISTINCT ins）OVER（PARTITION BY ord）或MIN（ins）OVER（PARTITION BY ord）结束为ins