从多列中获取最小值和次最小值的最简单方法是什么（MIN 函数不起作用）？

Question

Below table is from Teradata database

SELECT
  sku.Item_id,
  sku.Item_length,
  sku.Item_width,
  sku.Item_heigth,
FROM  Category_item sku

Item_id | Item_length | Item_width |Item_heigth
-------------------------------------------
104174        8            6           1
9482763       8            8           8
8434610       8            9           1
2109145       54           34          2
567106        41           41          2
2028731       3            4           3

The final result that I want is,

 Item_id | Item_length | Item_width |Item_heigth | MinValue | SecondMinVale
    ---------------------------------------------------------------------------
    104174        8            6           1          1            6
    9482763       8            8           8          8            8
    8434610       8            9           1          1            8
    2109145       54           34          2          2            34
    567106        41           41          2          2            41
    2028731       3            4           3          3            3

So I did a little research online, most of them used MIN function with a subquery or raw over by partition. Min function does not work for me, because I have three separate columns. I want to get minimum or least two values from the above three columns.

I tried to do subquery or CTE using Least function, but I am stuck with getting the second least value. I am not sure how to use raw over partition for this.

Any help regarding this will be highly appreciated.

Thank you so much!!

Answer 1

You can do it this way:

this below:

SELECT MIN(Item_caract)
      FROM (VALUES (Item_length),(Item_width),(Item_heigth)) a(Item_caract)

give you the minimum of all columns

and this:

(SELECT MIN(Item_caract)
      FROM (VALUES (Item_length),(Item_width),(Item_heigth)) a(Item_caract) where Item_caract >=
      (SELECT MIN(Item_caract)
      FROM (VALUES (Item_length),(Item_width),(Item_heigth)) a(Item_caract))

gives you the minimum that is bigger than the minimum above

Below is the full query:

select * ,

(SELECT MIN(Item_caract)
      FROM (VALUES (Item_length),(Item_width),(Item_heigth)) a(Item_caract))
   AS MinValue ,

(SELECT MIN(Item_caract)
      FROM (VALUES (Item_length),(Item_width),(Item_heigth)) a(Item_caract) where Item_caract >=
      (SELECT MIN(Item_caract)
      FROM (VALUES (Item_length),(Item_width),(Item_heigth)) a(Item_caract))
      )
   AS SecondMinVale 
from cte

Output:

104174  8   6   1   1   1
567106  41  41  2   2   2
2028731 3   4   3   3   3
2109145 54  34  2   2   2
8434610 8   9   1   1   1
9482763 8   8   8   8   8

Answer 2

Without set based analytic functions, your requirement is difficult. I might suggest unpivoting your data:

WITH cte AS (
    SELECT Item_id, Item_length AS item FROM Category_item UNION ALL
    SELECT Item_id, Item_width FROM Category_item UNION ALL
    SELECT Item_id, Item_height FROM Category_item
),
cte2 AS (
    SELECT Item_id, item,
        ROW_NUMBER() OVER (PARTITION BY Item_id ORDER BY item) rn
    FROM cte
)

SELECT
    t1.Item_id,
    t1.Item_length,
    t1.Item_width,
    t1.Item_height,
    MAX(CASE WHEN rn = 1 THEN t2.item END) AS MinValue,
    MAX(CASE WHEN rn = 2 THEN t2.item END) AS SecondMinValue
FROM Category_item t1
INNER JOIN cte2 t2
    ON t1.Item_id = t2.Item_id
WHERE
    t2.rn <= 2
GROUP BY
    t1.Item_id,
    t1.Item_length,
    t1.Item_width,
    t1.Item_height;

Note that the need to use such unpivoting operations might seriously imply that your table should be redesigned with the various measurements being stored per item across rows , rather than across columns.

Answer 3

If you actually got three columns you can simply apply this brute-force logic:

SELECT sku.*,
   Least(Item_length,Item_width,Item_height) AS MinValue,
   CASE
     WHEN Item_length BETWEEN Least(Item_width, Item_height) AND 
                           Greatest(Item_width, Item_height) 
     THEN Item_length
     WHEN Item_width BETWEEN Least(Item_length, Item_height) AND 
                           Greatest(Item_length, Item_height) 
     THEN Item_width
     WHEN Item_height BETWEEN Least(Item_width, Item_length) AND 
                           Greatest(Item_width, Item_length)
     THEN Item_height
   END AS SecondMinValue
FROM  Category_item sku

For more than three columns you need Tim's approach, this is a slight variation (if your TD release supports UNPIVOT) avoiding the join-back:

WITH cte AS 
 (
   SELECT up.*
     ,Row_Number() Over (PARTITION BY item_id ORDER BY val) AS rn
   FROM Category_item
   UNPIVOT
    ( val 
      FOR measure
      IN (Item_length AS 'l'
         ,Item_width  AS 'w'
         ,Item_height AS 'h') 
    ) AS up
 )  
SELECT
    Item_id,
    Max(CASE WHEN measure = 'l' THEN val END) AS item_length,
    Max(CASE WHEN measure = 'w' THEN val END) AS item_width,
    Max(CASE WHEN measure = 'h' THEN val END) AS item_height,
    Max(CASE WHEN rn = 1 THEN val END) AS MINVALUE,
    Max(CASE WHEN rn = 2 THEN val END) AS SecondMinValue
FROM cte
GROUP BY
    Item_id