忽略零值和空值列的多列之间的最小值

Question

我想从以下 5 列中创建一个名为 Market_Min 的新变量，其中：

Col 1 is 0
Col 2 is 100
Col 3 is 200
Col 4 is 150
Col 5 is NULL

在这种情况下，答案应该是 100。

Answer 1

使用 fn_Split 使用不同的方法

DECLARE @Temp AS TABLE (col1 int, col2 int,  col3 int, col4 int , col5 int)

INSERT INTO @Temp
    (col1 , col2 ,  col3 , col4 , col5 )
VALUES
    ( 0, 200, 100, 150,NULL)
SELECT
    *
    ,
    (
    SELECT 
        MIN(Value) 
    FROM 
        dbo.fn_Split
        (
          CAST(ISNULL(NULLIF(col1,0),99999) AS VARCHAR(10)) +','
        + CAST(ISNULL(NULLIF(col2,0),99999) AS VARCHAR(10)) +','
        + CAST(ISNULL(NULLIF(col3,0),99999) AS VARCHAR(10)) +','
        + CAST(ISNULL(NULLIF(col4,0),99999) AS VARCHAR(10)) +','
        + CAST(ISNULL(NULLIF(col5,0),99999) AS VARCHAR(10))
        ,','
        )
    ) AS MinNonZeroNonNullValue
FROM
    @Temp

您需要将 fn_Split 函数添加到服务器。可以下载Here

Answer 2

这是使用union all 获取整个表格的最小值的一个选项：

select min(combinedcol)
from (
    select col1 combinedcol from yourtable union all
    select col2 from yourtable union all
    select col3 from yourtable union all
    select col4 from yourtable union all
    select col5 from yourtable
) t
where coalesce(combinedcol,0) > 0

---

根据 cmets 编辑基础

如果您需要每行的最小值，您可以在子查询中引入row_number 和group by：

select rn, min(combinedcol)
from (
    select row_number() over (order by (select null)) rn, col1 combinedcol from yourtable union all
    select row_number() over (order by (select null)) rn, col2 from yourtable union all
    select row_number() over (order by (select null)) rn, col3 from yourtable union all
    select row_number() over (order by (select null)) rn, col4 from yourtable union all
    select row_number() over (order by (select null)) rn, col5 from yourtable
) t
where coalesce(combinedcol,0) > 0
group by rn

• Condensed SQL Fiddle Demo