TSQL選擇最小和最大的行分組

Question

的時候可以說我有一個包含許多類似的許多行的表：

ID  Range   Range_begining  Profit 
---------------------------------------------------- 
1 (100-150)     100   -20 
2 (200-250)     200   40.2 
3 (100-150)     100   100 
4 (450-500)     450   -90 
... 

我做這樣一個簡單的查詢：

SELECT max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    FROM 
     Orders 
    GROUP BY 
     Range_begining 

在此之後運行查詢時，我得到的結果是這樣的：

Range  Count  AVG Profit 
------------------------------------ 
(100-150)  2    40 
(200-250)  1    40.2 
(450-500)  1    -90 
... 

很簡單:)

我現在需要做的是用最小的和最大的利潤，其中數是選擇行大於10（這是一個參數）

我能得到最小值與此：

SELECT TOP 1 [Range], [AVG Profit] FROM (
    SELECT max([Range]) AS 'Range' 
     , count(ID) AS 'Count' 
     , round(avg([Profit]), 2) AS 'AVG Profit' 
     FROM 
      Orders 
     GROUP BY 
      Range_begining) X 
WHERE 
    [Count]>10 
ORDER BY 
    [AVG Profit] ASC --or DESC if I want max profit 

我正在考慮使用ORDER BY DESC爲上述查詢做一個UNION，但這不是最好的解決方案。

我需要做的事情：
選擇2行：一個最小值，一個最大AVG利用Range進行分組時獲利。

編輯： 如果我加2移動列到我的主數據表是這樣的：

ID  Range   Range_begining  Profit  OrderDate  Company 
--------------------------------------------------------------------------------- 
1 (100-150)     100   -20  2012-01-02   1 
2 (200-250)     200   40.2  2012-03-22   0 
3 (100-150)     100   100  2012-02-05   0 
4 (450-500)     450   -90  2012-05-12   1 
... 

，然後嘗試添加2個條件是這樣的：

; with ordering as (
    SELECT max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    , row_number() over (order by avg([Profit])) rn_min 
    , row_number() over (order by avg([Profit]) desc) rn_max 
    FROM 
     Orders 
    GROUP BY 
     Range_begining 
    HAVING COUNT(ID) > 10 
    AND [Company]=@company 
    AND (@from= '' OR [OrderDate]>=@from) 
    AND (@to= '' OR [OrderDate]<=@to) 
) 
select [range], [count], [avg profit] 
    from ordering 
where (rn_max = 1 or rn_min = 1) 

我得到一個錯誤，因爲[公司]和[OrderDate]

在HAVING子句中無效，因爲它不包含在 集合函數或GROUP BY子句中。

我該如何解決這個問題？

EDIT2 明白了！

; with ordering as (
    SELECT max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    , row_number() over (order by avg([Profit])) rn_min 
    , row_number() over (order by avg([Profit]) desc) rn_max 
    FROM 
     Orders 
    WHERE 
    [Company]=@company 
    AND (@from= '' OR [OrderDate]>=@from) 
    AND (@to= '' OR [OrderDate]<=@to) 
    GROUP BY 
     Range_begining 
    HAVING COUNT(ID) > 10 
) 
select [range], [count], [avg profit] 
    from ordering 
where (rn_max = 1 or rn_min = 1) 

編輯3 我可以用描述返回另一列是這樣的：

Range  AVG Profit    Description 
------------------------------------------------- 
(200-250)   40.2   Max profit here 
(450-500)   -90  Min profit, well done 

EDIT 4 快速的答案（基於@NikolaMarkovinović答案）：

; with ordering as (
    SELECT max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    , row_number() over (order by avg([Profit])) rn_min 
    , row_number() over (order by avg([Profit]) desc) rn_max 
    FROM 
     Orders 
    WHERE 
    [Company]=@company 
    AND (@from= '' OR [OrderDate]>=@from) 
    AND (@to= '' OR [OrderDate]<=@to) 
    GROUP BY 
     Range_begining 
    HAVING COUNT(ID) > 10 
) 
    SELECT 
    CASE WHEN rn_max=1 THEN 'This is max' ELSE 'Min' END AS 'Description' 
    ,[range] 
    ,[count] 
    ,[avg profit] 
    FROM ordering 
    WHERE (rn_max = 1 or rn_min = 1) 

Answer 1

你可能會使用window functions一次做到這一點：

; with ordering as (
    SELECT max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    , row_number() over (order by avg([Profit])) rn_min 
    , row_number() over (order by avg([Profit]) desc) rn_max 
    FROM 
     Orders 
    GROUP BY 
     Range_begining 
    HAVING COUNT(ID) > 10 
) 
select [range], [count], [avg profit], 
     case when rn_max = 1 
      then 'Max profit' 
      else 'Min profit' 
     end Description 
    from ordering 
where (rn_max = 1 or rn_min = 1) 

And here is Sql Fiddle example。

Answer 2

這是SQLFiddle example。在你的最後一個問題查詢，您可以使用具有代替嵌套查詢：

select * from 
(SELECT TOP 1 
    max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    FROM 
     Orders 
    GROUP BY 
     Range_begining 
    having count(id)>10 
    order by round(avg([Profit]), 2) ASC 
) a 
union all 
select * from 
(
SELECT TOP 1 
    max([Range]) AS 'Range' 
    , count(ID) AS 'Count' 
    , round(avg([Profit]), 2) AS 'AVG Profit' 
    FROM 
     Orders 
    GROUP BY 
     Range_begining 
    having count(id)>10 
    order by round(avg([Profit]), 2) desc 
)b