SQL查询返回一个列值，其中另一个列是每个序列号的最小值答案

【问题标题】：SQL query to return a column value where another column is minimum per serial numberSQL查询返回一个列值，其中另一个列是每个序列号的最小值
【发布时间】：2014-09-15 18:30:27
【问题描述】：

我有一个如下表，并希望通过查询为我提供一个新列：第一个站点。表：

--------------------------------------------------------------
Serial_No | Production_Date |   Production_Site |   First_Site
--------------------------------------------------------------
001       | 11 June 2013    |   SiteA           |   SiteA
--------------------------------------------------------------
001       | 12 June 2013    |   SiteB           |   SiteA
--------------------------------------------------------------
001       | 12 June 2013    |   SiteC           |   SiteA
--------------------------------------------------------------
001       | 10 June 2013    |   SiteA           |   SiteA
--------------------------------------------------------------
002       | 14 July 2013    |   SiteB           |   SiteA
--------------------------------------------------------------
002       | 14 July 2013    |   SiteC           |   SiteA
--------------------------------------------------------------
002       | 9 May 2013      |   SiteA           |   SiteA
--------------------------------------------------------------
003       | 5 June 2013     |   SiteB           |   SiteB
--------------------------------------------------------------
003       | 19 July 2013    |   SiteC           |   SiteB
--------------------------------------------------------------

First_Site 列读取每个序列号，找到该序列号的最早生产日期，并输出对应于最早日期的站点。

我尝试了以下查询，但他们没有给我想要的结果：

SELECT
Case When ( Production_Date ) = ( MIN( Production_Date ) 
OVER (PARTITION BY ( Serial_No)) ) 
Then (Production_Site) Else 'Null') End AS First_Site
FROM Table

这为每个序列号的 Production_Date 最小的行提供了正确的值，但为序列号的其他行提供了空值：

--------------------------------------------------------------
Serial_No | Production_Date |   Production_Site |   First_Site
--------------------------------------------------------------
001       | 11 June 2013    |   SiteA           |   Null
--------------------------------------------------------------
001       | 12 June 2013    |   SiteB           |   Null
--------------------------------------------------------------
001       | 12 June 2013    |   SiteC           |   Null
--------------------------------------------------------------
001       | 10 June 2013    |   SiteA           |   SiteA
--------------------------------------------------------------
002       | 14 July 2013    |   SiteB           |   Null
--------------------------------------------------------------
002       | 14 July 2013    |   SiteC           |   Null
--------------------------------------------------------------
002       | 9 May 2013      |   SiteA           |   SiteA
--------------------------------------------------------------
003       | 5 June 2013     |   SiteB           |   SiteB
--------------------------------------------------------------
003       | 19 July 2013    |   SiteC           |   Null
--------------------------------------------------------------

这不是我想要的。我的猜测是我必须更改案例陈述的 ELSE 部分。有人可以帮忙吗？

【问题讨论】：

您使用的是哪个 DBMS？甲骨文？ Postgres？
我正在使用 Oracle DBMS

标签： sql case min

【解决方案1】：

这似乎行得通。

CREATE TABLE #table (Serial_No int, Production_Date datetime, Production_Site varchar(50), First_Site varchar(50))

INSERT INTO #table VALUES(001, '11 June 2013', 'SiteA', 'SiteA')
INSERT INTO #table VALUES(001, '12 June 2013', 'SiteB', 'SiteA')
INSERT INTO #table VALUES(001, '12 June 2013', 'SiteC', 'SiteA')
INSERT INTO #table VALUES(001, '10 June 2013', 'SiteA', 'SiteA')
INSERT INTO #table VALUES(002, '14 July 2013', 'SiteB', 'SiteA')
INSERT INTO #table VALUES(002, '14 July 2013', 'SiteC', 'SiteA')
INSERT INTO #table VALUES(002, '9 May 2013', 'SiteA', 'SiteA')
INSERT INTO #table VALUES(003, '5 June 2013', 'SiteB', 'SiteB')
INSERT INTO #table VALUES(003, '19 July 2013', 'SiteC', 'SiteB')

SELECT 
    main.*
    ,(
        SELECT MIN(Production_Date) 
        FROM #table AS sub 
        WHERE main.Serial_No = sub.Serial_No
    ) AS Earliest_Production_Date
FROM
    #table as main

但我想这看起来更好：

SELECT 
    main.*
    ,MIN(Production_Date) OVER (PARTITION BY Serial_No)
FROM
    #table as main

或者，如果您想从最早的生产中获取更多数据：

WITH cte AS (
    SELECT 
        main.*
        ,MIN(Production_Date) OVER (PARTITION BY Serial_No) AS Earliest_Production_Date
    FROM
        #table as main
)
SELECT
    cte.*
    ,t.First_Site AS Earliest_First_Site
FROM
    cte
    INNER JOIN #table AS t ON 
        t.Serial_No = cte.Serial_No 
        and t.Production_Date = cte.Earliest_Production_Date

【讨论】：

【解决方案2】：

create table First_Site (serial_no varchar(20), production_date varchar(20), production_site varchar(20), first_site varchar(22))

insert into First_Site select '001', '11 June 2013', 'SiteA', 'SiteA'
insert into First_Site select '001', '12 June 2013', 'SiteB', 'SiteA'
insert into First_Site select '001', '12 June 2013', 'SiteC', 'SiteA'
insert into First_Site select '001', '10 June 2013', 'SiteA', 'SiteA'
insert into First_Site select '002', '14 July 2013', 'SiteB', 'SiteA'
insert into First_Site select '002', '14 July 2013', 'SiteC', 'SiteA'
insert into First_Site select '002', '9 May 2013', 'SiteA', 'SiteA'
insert into First_Site select '003', '5 June 2013', 'SiteB', 'SiteB'
insert into First_Site select '003', '19 June 2013', 'SiteC', 'SiteB'

SELECT fs.serial_no, fs.production_date, A.dt, fs.production_site, fs.first_site FROM first_site fs
JOIN 
(
    SELECT serial_no, MIN(CONVERT(DATETIME, production_date)) dt , first_site
    FROM first_site 
    GROUP BY serial_no, first_site
)A ON a.dt = fs.production_date 
AND A.first_site = fs.first_site

【讨论】：

【解决方案3】：

这是一个可行的解决方案：

SELECT main.*, sub.production_site as first_site 
FROM t main INNER JOIN t sub ON main.serial_no = sub.serial_no
WHERE NOT EXISTS 
  (select * from t 
   where serial_no = sub.serial_no and production_date < sub.production_date);

http://sqlfiddle.com/#!2/79d962/4

此代码的唯一问题是，如果您可以在同一日期拥有多个具有相同序列号的站点，您最终将得到两条记录。

【讨论】：