SQL Server 将行转置为列

Question

我需要使用SQL Server将行转列以显示不同维度的数据或满足表格格式要求。

从这个查询和这个返回开始

SELECT 
  _p,
  _t,
  _x,
 COUNT (*) _q 
FROM
    [dbo].[subset_data] 
WHERE
    CONVERT ( VARCHAR ( 10 ), data, 120 ) = CONVERT ( VARCHAR ( 10 ), GETDATE( ), 120 ) 
    GROUP BY
    _p,
    _t,
    _x;


+--------+--------------+----------+--------+
| _p     | _t           |   _x     |   _q   |
+--------+--------------+----------+--------+
| 001    | 06:00-12:00  |   15     |    1   |
| 001    | 12:00-18:00  |   15     |    1   |
| 008    | 06:00-12:00  |    5     |    2   |
| 009    | 12:00-18:00  |    6     |    1   |
| 010    | 06:00-12:00  |   12     |    1   |
+--------+--------------+----------+--------+

我必须得到这份回报

+--------+--------------+-------------+-------------+-------------+------------+
| _p     | 00:00-06:00  | 06:00-12:00 | 12:00-18:00 | 18:00-23:59 |    _x      |
+--------+--------------+-------------+-------------+-------------+------------+
| 001    |      0       |      1      |      1      |      0      |    15      |
| 008    |      0       |      2      |      0      |      0      |     5      |
| 009    |      0       |      0      |      1      |      0      |     6      |
| 010    |      0       |      1      |      0      |      0      |    12      |
+--------+--------------+-------------+-------------+-------------+------------+

列 _t 的行值要转换为列，对于 _p 列的每个值，我必须从 _q 列检索行的值。

你能帮我吗？

-- INIT database
CREATE TABLE subset_data (
   data DATETIME,
  _p VARCHAR(100),
  _t VARCHAR(100),
  _x VARCHAR(100),
);

INSERT INTO subset_data(data, _p, _t, _x) VALUES (GETDATE(), '001', '06:00-12:00', '15');
INSERT INTO subset_data(data, _p, _t, _x) VALUES (GETDATE(), '001', '12:00-18:00', '15' );
INSERT INTO subset_data(data, _p, _t, _x) VALUES (GETDATE(), '008', '06:00-12:00', '5' );
INSERT INTO subset_data(data, _p, _t, _x) VALUES (GETDATE(), '008', '06:00-12:00', '5' );
INSERT INTO subset_data(data, _p, _t, _x) VALUES (GETDATE(), '009', '12:00-18:00', '6');
INSERT INTO subset_data(data, _p, _t, _x) VALUES (GETDATE(), '010', '06:00-12:00', '12');

SQLFiddle

我尝试使用

PIVOT

如下：

select  
  _p, _t, _x
from
(
  select _p
  from [dbo].[subset_data] 
) d
pivot
(
  _p
  for _p in (_p, _t, _x)
) piv;

错误

> Msg 156, Level 15, State 1, Server IT000001861251, Procedure , Line 0
Incorrect syntax near the keyword 'for'.
> [42000] [Microsoft][ODBC Driver 17 for SQL Server][SQL Server]Incorrect syntax near the keyword 'for'. (156)

Answer 1

您所做的尝试有很多的缺陷：

1. 您的派生表

   d

   仅公开列

   _p

   ，因此

   _t

   和

   _x

   不可访问。
2. 您的

   PIVOT

   缺少聚合函数。
3. 您正在尝试检查

   _p

   的值，但您想要检查

   _t

4. 您正在检查该列是否 (

   _p) has the value 

   '_p'

   , 

   '_t'

   , or 

   '_x'

    but those values never appear in that column, nor the column you actually want to pivot on (

   _t`)。
5. 您的

   IN

   子句中有一个尾随逗号。

建议的重复内容，在 sql server 中高效地将行转换为列，向您展示了如何正确使用

PIVOT

，其格式为：

SELECT <Columns List from Pivoted derived table>
FROM (SELECT <Columns List>
      FROM <Table Name> --Your Table) <Alias>
PIVOT (<Aggregate Expression>
       FOR <Column to PIVOT on> IN (<Delimited list of Values to PIVOT on>)) <Pivot Alias>;

对于您的数据，这会产生以下结果：

SELECT _p,
       [00:00-06:00],
       [06:00-12:00],
       [12:00-18:00],
       [18:00-23:59],
       _x
FROM (SELECT _p,
             _t,
             _x
        FROM [dbo].[subset_data]) d
PIVOT (COUNT(_t)
       FOR _t IN ([00:00-06:00],[06:00-12:00],[12:00-18:00],[18:00-23:59])) piv
ORDER BY _p;

不过，就我个人而言，一如既往，我建议条件聚合；语法不那么“笨重”，限制更少，并且（更）可以转移到其他方言：

SELECT _p,
       COUNT(CASE _t WHEN '00:00-06:00' THEN 1 END) AS [00:00-06:00],
       COUNT(CASE _t WHEN '06:00-12:00' THEN 1 END) AS [06:00-12:00],
       COUNT(CASE _t WHEN '12:00-18:00' THEN 1 END) AS [12:00-18:00],
       COUNT(CASE _t WHEN '18:00-23:59' THEN 1 END) AS [18:00-23:59],
       _x
FROM dbo.subset_data sd
GROUP BY _p,
         _x
ORDER BY _p;