Question

我总共有10个小组。分组编号（列A）在列Type中的每个不同值处重新开始。我希望将所有类型的10个组除以100，然后按每个组中的条目数（行数）得到每组的结果数。

列A + Type是我的组标识符，列B是我想要的最终结果。

我尝试了一些基于嵌套查询的变体方程式，其中的计数按A和type分组，但我并没有理解如何将它们整合在一起。

ID  A   B   TYPE

01  1   10  APPLE
02  2   5   APPLE
03  2   5   APPLE
04  3   2.5 APPLE
05  3   2.5 APPLE
06  3   2.5 APPLE
07  3   2.5 APPLE
08  4   5   APPLE
09  4   5   APPLE
10  5   10  APPLE
11  1   5   ORANGE
12  1   5   ORANGE
13  2   10  ORANGE
14  3   5   ORANGE
15  3   5   ORANGE
16  4   2.5 ORANGE
17  4   2.5 ORANGE
18  4   2.5 ORANGE
19  4   2.5 ORANGE
20  5   10  ORANGE

Answer 1

<强> EDITED

修复附加行

21  5   **  ORANGE

SQL Fiddle Demo

WITH subGroup as (
   SELECT [TYPE], [A], count([TYPE]) as total
   FROM Table1
   GROUP BY [TYPE], [A]
),
totalGroup as (
    SELECT count(DISTINCT [TYPE] + CAST([A] as varchar(10))) as total
    FROM Table1        
)
SELECT t.ID, t.[A], (100.0 / tg.total / s.total) as [B], t.[TYPE]
FROM Table1 t
inner join subGroup s
  on t.[Type] = s.[Type]
 and t.[A] = s.[A]
, totalGroup tg

<强>输出

| ID | A |   B |   TYPE |
|----|---|-----|--------|
|  1 | 1 |  10 |  APPLE |
|  2 | 2 |   5 |  APPLE |
|  3 | 2 |   5 |  APPLE |
|  4 | 3 | 2.5 |  APPLE |
|  5 | 3 | 2.5 |  APPLE |
|  6 | 3 | 2.5 |  APPLE |
|  7 | 3 | 2.5 |  APPLE |
|  8 | 4 |   5 |  APPLE |
|  9 | 4 |   5 |  APPLE |
| 10 | 5 |  10 |  APPLE |
| 11 | 1 |   5 | ORANGE |
| 12 | 1 |   5 | ORANGE |
| 13 | 2 |  10 | ORANGE |
| 14 | 3 |   5 | ORANGE |
| 15 | 3 |   5 | ORANGE |
| 16 | 4 | 2.5 | ORANGE |
| 17 | 4 | 2.5 | ORANGE |
| 18 | 4 | 2.5 | ORANGE |
| 19 | 4 | 2.5 | ORANGE |
| 20 | 5 |   5 | ORANGE |
| 21 | 5 |   5 | ORANGE |

Answer 2

希望此查询有帮助

Select At.A, B = CAST(tot as float) / Cnt, At.Type 
From Test AT
Inner Join 
    (Select A, Type, Cnt = Count(*) From Test
    Group By A, Type) BT On AT.A = BT.A and AT.Type = BT.Type
Inner Join 
    (Select Type, Tot = Count(A) From Test
    Group By Type) CT On CT.Type = AT.Type`

Answer 3

如果您的SQL Server是2008 +：

，您可以使用窗口函数来执行此操作

SELECT
    ID,
    A,
    (COUNT(*) OVER (PARTITION BY TYPE))/(COUNT(*) OVER (PARTITION BY A,TYPE)) as B,
    TYPE
FROM <table>
order by TYPE, A asc

要添加的编辑：我真的不明白“除以100 ......”部分，但这里的数学符合您想要的结果...所以我认为它的效果相同，对吗？

将每个分组中的行数除以单个数字

3 个答案: