SQL 将特定列中的所有数据行转换为列，并将另一列中的值映射到新列

Question

来源TABLE 看起来像这样：

h1_ind = 1 表示包含列名的行。

Source                              
    col1    col2        col3     col4       h1_ind  table_sheet y   w
    Route   CNG/Diesel  Freq     Weekly Miles   1   Summary   2021  Week 1
    a        5          B        6              2   Summary   2021  Week 1
    b        1          1        1              3   Summary   2021  Week 1
    c        5          B        6              4   Summary   2021  Week 1  
    d        1          1        1              5   Summary   2021  Week 1
    Route   CNG/Diesel  Freq     Weekly Miles   1   Summary   2021  Week 2
    a        5          B        6              2   Summary   2021  Week 2
    b        1          1        1              3   Summary   2021  Week 2
    c        5          B        6              4   Summary   2021  Week 2  
    d        1          1        1              5   Summary   2021  Week 2

然后我转换成这样：（仅限第 1 周）

Data                
IN      column           table_sheet     y      w     value h1_ind
    col1    Route        Summary        2021    Week 1  a   2
    col2    CNG/Diesel   Summary        2021    Week 1  5   3
    col3    Freq         Summary        2021    Week 1  B   4
    col4    Weekly Miles Summary        2021    Week 1  6   5
    col1    Route        Summary        2021    Week 1  b   2
    col2    CNG          Summary        2021    Week 1  1   3
    col3    Freq         Summary        2021    Week 1  1   4
    col4    Weekly Miles Summary        2021    Week 1  1   5
    col1    Route        Summary        2021    Week 1  c   2
    col2    CNG/Diesel   Summary        2021    Week 1  5   3
    col3    Freq         Summary        2021    Week 1  B   4
    col4    Weekly Miles Summary        2021    Week 1  6   5
    col1    Route        Summary        2021    Week 1  d   2
    col2    CNG          Summary        2021    Week 1  1   3
    col3    Freq         Summary        2021    Week 1  1   4
    col4    Weekly Miles Summary        2021    Week 1  1   5

我希望能够按照以下要求呈现。
（仅限第 1 周）

Desired output:
table_sheet    y      w      Route CNG/Diesel  Freq  Weekly Miles
Summary       2021   Week 1   a       5         B     6
Summary       2021   Week 1   b       1         1     1
Summary       2021   Week 1   c       5         B     6
Summary       2021   Week 1   d       1         1     1

Desired Mapping

我该怎么做？ Pivot 只给我一行 (MAX, MIN) 我需要显示所有行。

Answer 1

可以使用PIVOT，关键是要确保h1_ind也是分组的一部分，而不是被聚合。

我们使用窗口聚合来获取每组值的 header 名称
使用 CROSS APPLY
然后使用 PIVOT 聚合备份。

WITH Headers AS (
    SELECT *,
      Col1Header = MAX(CASE WHEN h1_ind = '1' THEN col1 END) OVER (PARTITION BY table_sheet, y, w),
      Col2Header = MAX(CASE WHEN h1_ind = '1' THEN col2 END) OVER (PARTITION BY table_sheet, y, w),
      Col3Header = MAX(CASE WHEN h1_ind = '1' THEN col3 END) OVER (PARTITION BY table_sheet, y, w),
      Col4Header = MAX(CASE WHEN h1_ind = '1' THEN col4 END) OVER (PARTITION BY table_sheet, y, w)
    FROM YourTable t
),
Unpivoted AS (
    SELECT
      t.table_sheet,
      t.y,
      t.w,
      t.h1_ind,
      v.Header,
      v.Value
    FROM Headers t
    CROSS APPLY (VALUES
        (Col1Header, Col1),
        (Col2Header, Col2),
        (Col3Header, Col3),
        (Col4Header, Col4)
    ) v(Header, Value)
    WHERE h1_ind > 1
)
SELECT *
FROM Unpivoted
PIVOT (
    MAX(Value) FOR Header IN
    (Route, [CNG/Diesel], Freq, [Weekly Miles])
) pvt;

db<>fiddle

如果需要，您也可以在此处使用 MAX(CASE 条件聚合，而不是 PIVOT。

WITH ...
SELECT
  t.table_sheet,
  t.y,
  t.w,
  Route          = MAX(CASE WHEN v.Header = 'Route'        THEN v.Value END),
  [CNG/Diesel]   = MAX(CASE WHEN v.Header = 'CNG/Diesel'   THEN v.Value END),
  Freq           = MAX(CASE WHEN v.Header = 'Freq'         THEN v.Value END),
  [Weekly Miles] = MAX(CASE WHEN v.Header = 'Weekly Miles' THEN v.Value END)
FROM Unpivoted
GROUP BY
  t.table_sheet,
  t.y,
  t.w
  t.h1_ind;

SQL 将特定列中的所有数据行转换为列，并将另一列中的值映射到新列

SQL Convert all rows of data in a specific column to columns and mapping values in another column to the new columns

sql

tsql

pivot

pentaho

pentaho-spoon