需要关于 PIVOT 是否是 SQL 服务器中的正确工具的建议

Question

我的数据是这样的：

| ACCNT | AMOUNT | TYPE | YEAR |
|-------|--------|------|------|
|  458  | 168.80 | A    | 2014 |
|  458  | 282.77 | A    | 2015 |
|  458  | 64.70  | B    | 2015 |
|  458  | 91.25  | A    | 2016 |
|  458  | 398.00 | B    | 2016 |
|  458  | 104.45 | C    | 2016 |
|  927  | 445.00 | B    | 2014 |
|  927  | 25.00  | A    | 2015 |
|  927  | 10.00  | NULL | 2015 |
|  927  | 132.00 | A    | 2016 |
|  927  | 381.40 | B    | 2016 |
|  927  | 210.00 | C    | 2016 |
...

Accnt 可能没有每个 Year 的值，并且 Type 偶尔会为 NULL。我想每个账户只有一行，并在年份标题下合并金额和类型列，即让它看起来像这样：

| ACCNT |   2014    |       2015         |           2016             |
|-------|-----------|--------------------|----------------------------|
|  458  | 168.80,A  | 282.77,A;64.70,B   | 91.25,A;398.00,B;104.45,C  |  
|  927  | 445.00,B  | 25.00,A;10.00,NULL | 132.00,A;381.40,B;210.00,C |
...

我可以在[年]...

SELECT * FROM 
 (SELECT [Accnt], [Amount], [Type], [Year] FROM data_table) 
 AS Source
PIVOT
(
 MAX([Amount]) FOR [Year] IN ([2014],[2015],[2016])
) AS PVT

...但每个帐户仍然有多行。

以下更接近我正在寻找的格式...

SELECT Accnt, [1] AS '2014', [2] AS '2015', [3] AS '2016'
FROM (SELECT ROW_NUMBER() OVER (PARTITION BY Accnt ORDER BY [Year] Asc) AS AccountID, 
Accnt, CONVERT(nvarchar,Amount) + ',' + [Type]  AS Amount_And_Type
FROM  data_table) PIVOT (MAX(Amount_And_Type) 
FOR AccountID IN ([1], [2], [3])) AS data_pvt

...但它只是给我每个帐户的前三行：

| ACCNT |   2014    |   2015   |  2016   |
|-------|-----------|----------|---------|
|  458  | 168.80,A  | 282.77,A | 67.40,B |  
|  927  | 445.00,B  | 25.00,A  | NULL    |

我想用 Pivot 做的事情是否可行？

Answer 1

您可以通过以下方式查询：

;with cte as (
    select Accnt, iif([2014] is null,null, concat([2014],',',[Type])) as [2014a]
        ,iif([2015] is null,null, concat([2015],',',[Type])) as [2015a]
        ,iif([2016] is null,null, concat([2016],',',[Type])) as [2016a]
     from youracct
    pivot(max(amount) for [year] in ([2014],[2015],[2016])) p
) 
select c.Accnt, [2014]= stuff(( Select ';'+[2014a] from cte where Accnt = c.Accnt for xml path('')),1,1,'')    
,[2015]=stuff(( Select ';'+[2015a] from cte where Accnt = c.Accnt for xml path('')),1,1,'')  
,[2016]=stuff(( Select ';'+[2016a] from cte where Accnt = c.Accnt for xml path('')),1,1,'')  
from cte c
group by c.Accnt

但是这种方法的动态 sql 有点复杂，但我们可以做到..

输出如下：

+-------+---------+-----------------+------------------------+
| Accnt |  2014   |      2015       |          2016          |
+-------+---------+-----------------+------------------------+
|   458 | 168.8,A | 282.77,A;64.7,B | 91.25,A;398,B;104.45,C |
|   927 | 445,B   | 10,;25,A        | 132,A;381.4,B;210,C    |
+-------+---------+-----------------+------------------------+

需要关于 PIVOT 是否是 SQL 服务器中的正确工具的建议

Need advice on whether PIVOT is the correct tool in SQL Server

sql-server

sql-server-2008-r2

pivot

row-number