分區OVER場另一場

我有這個表分區OVER場另一場

ID UPC Sales Date  
1 333 10  1/1/2015 
1 222 20  1/1/2015 
1 111 30  1/1/2015 
1 444 10  2/1/2015 
1 555 20  2/1/2015 
2 333 20  1/1/2015 
2 222 50  1/1/2015 
2 111 30  1/1/2015 
2 444 20  2/1/2015 
2 555 20  2/1/2015

而且希望這個輸出

我已經試過這

SELECT ID, AVG(Sales) OVER (Partition BY Date) as basket_size 
FROM Transactions 
GROUP BY ID

但我得到這個錯誤分組時：

Column 'Transactions.Date' is invalid in the select list because it is not contained in either an aggregate function or the GROUP BY clause.

想法？

來源

2015-04-14 user2502836

下面會給你想要的結果：

DECLARE @t TABLE(ID INT,UPC INT, Sales INT, [Date] DATE); 
INSERT INTO @t(ID,UPC,Sales,[Date]) 
VALUES(1,333,10,'2015-01-01'), 
(1,222,20,'2015-01-01'), 
(1,111,30,'2015-01-01'), 
(1,444,10,'2015-01-02'), 
(1,555,20,'2015-01-02'), 
(2,333,20,'2015-01-01'), 
(2,222,50,'2015-01-01'), 
(2,111,30,'2015-01-01'), 
(2,444,20,'2015-01-02'), 
(2,555,20,'2015-01-02'); 

SELECT 
    t1.ID, 
    t1.[SUM(Sales)], 
    t2.basket_size AS [AVG(Sales)], 
    t1.[COUNT(DISTINCT Date)] 
FROM 
    (
     SELECT 
      ID, 
      SUM(Sales) AS [SUM(Sales)], 
      COUNT(DISTINCT [Date]) AS [COUNT(DISTINCT Date)] 
     FROM 
      @t 
     GROUP BY 
      ID 
    ) AS t1 
    INNER JOIN(
     SELECT 
      ID, 
      AVG([SUM(Sales)]) AS basket_size 
     FROM 
      (
       SELECT 
        ID, 
        [Date], 
        SUM(Sales) AS [SUM(Sales)] 
       FROM 
        @t 
       GROUP BY 
        ID, 
        [Date] 
      ) AS tt 
     GROUP BY  
      ID 
    ) AS t2 ON 
     t1.ID=t2.ID

來源

2015-04-14 15:28:19

這工作。謝謝TT – user2502836

@ user2502836思考了一會兒，我想出了一個更簡單的查詢:)：SELECT ID，SUM（Sales）AS sum，SUM（Sales）/ COUNT（DISTINCT [Date]）AS [avg] ，COUNT（DISTINCT [Date]）AS [count] FROM @t GROUP BY ID;'。我相信這對你的要求也是正確的，只是簡單得多。 –

這可以通過使用OUTER APPLY完成。試試這個，

CREATE TABLE Transactions(ID INT, UPC INT, Sales INT, [Date] Date) 

INSERT INTO Transactions(ID, UPC, Sales, [Date]) 
VALUES 
(1, 333, 10,  '1/1/2015'), 
(1, 222, 20,  '1/1/2015'), 
(1, 111, 30,  '1/1/2015'), 
(1, 444, 10,  '2/1/2015'), 
(1, 555, 20,  '2/1/2015'), 
(2, 333, 20,  '1/1/2015'), 
(2, 222, 50,  '1/1/2015'), 
(2, 111, 30,  '1/1/2015'), 
(2, 444, 20,  '2/1/2015'), 
(2, 555, 20,  '2/1/2015') 

SELECT ID, SalesSum AS [Sum], SalesSum/[COUNT] AS AVG, [COUNT] AS [Count] FROM (
SELECT ID, Sum(Sales) OVER (PARTITION BY ID) AS SalesSum, [COUNT], 
ROW_NUMBER() OVER (PARTITION BY ID ORDER BY ID) RN 
FROM Transactions 
OUTER APPLY 
(
    SELECT COUNT(DISTINCT [Date]) [COUNT] FROM Transactions 
) AS OUTR 
) TMP WHERE RN = 1

Sql Fiddle Demo

來源

2015-04-14 15:38:09

分區OVER場另一場

回答

相關問題