根據某些字段（email_id，role_id和dob）和基於條件（creation_date）刪除重複記錄

Question

我有一個名爲Customer的表，它包含基於某些字段的重複記錄。
顧客表：
的cust_id
EMAIL_ID
ROLE_ID
DOB
CREATION_DATE

數據在客戶表是如下：

cust_id email_id   role_id  dob  creation_date 
1  abc@abc.com   5  4/2/1966  17/09/2016 
2  abc@abc.com   5  4/2/1966  20/09/2016 
3  xyz@xyz.com   5  15/2/1991  18/09/2016 
4  xyz@xyz.com   5  15/2/1991  21/09/2016 
5  pqr@pqr.com   5  16/2/1985  30/09/2016 
6  pqr@pqr.com   5  16/2/1985  05/11/2016 
7  pqr@pqr.com   5  16/2/1985  04/11/2016 

的EMAIL_ID，ROLE_ID和DOB是相同（重複）2個或更多記錄，如上表所示。

我想不同的查詢，這將導致如下：

cust_id email_id   role_id  dob  creation_date 
1  abc@abc.com   5  4/2/1966  17/09/2016  
3  xyz@xyz.com   5  15/2/1991  18/09/2016 
5  pqr@pqr.com   5  16/2/1985  30/09/2016  

也就是說，除去基於EMAIL_ID，ROLE_ID和出生日期，其CREATION_DATE小於其他重複記錄重複的記錄。

cust_id email_id   role_id  dob  creation_date 
2  abc@abc.com   5  4/2/1966  20/09/2016  
4  xyz@xyz.com   5  15/2/1991  21/09/2016 
6  pqr@pqr.com   5  16/2/1985  05/11/2016 

也就是說，除去基於EMAIL_ID，ROLE_ID和出生日期，其CREATION_DATE比其他重複的記錄更多的重複記錄。

編輯：對上述問題的反問題。

現在，當我在兩個名爲Customer和Individual的表上加入連接時，如何獲得與上面相同的所需結果。
客戶表：
CUST_ID
EMAIL_ID
ROLE_ID
individaul_id（外鍵）
CREATION_DATE

Individaul表：
使用下面的查詢individaul_id
DOB

：

SELECT c.email_id,c.role_id,i.dob FROM CUSTOMER c 
JOIN INDIVIDUAL i on c.individaul_id=i.individaul_id  
GROUP BY c.email_id,c.role_id,i.dob  
Having count(*) >=2 

我使用MSSQL Server 2012的數據庫。
非常感謝提前。

Answer 1

我使用了@navintb回答的查詢，並將其修改如下，以刪除重複的結果以獲取所需的輸出。

SELECT max(cust_id),c.email_id,c.role_id,i.dob,max(creation_date) FROM 
CUSTOMER c 
JOIN INDIVIDUAL i on c.individual_id=i.individual_id 
GROUP BY c.email_id,c.role_id,i.dob 
Having count(*) >=2 

AND，

SELECT min(cust_id),c.email_id,c.role_id,i.dob,min(creation_date) FROM 
CUSTOMER c 
JOIN INDIVIDUAL i on c.individual_id=i.individual_id 
GROUP BY c.email_id,c.role_id,i.dob 
Having count(*) >=2 

Answer 2

使用MIN和MAX函數

select min(cust_id),email_id,role_id,dob,min(creation_date) from customer group by email_id,role_id,dob; 

select max(cust_id),email_id,role_id,dob,max(creation_date) from customer group by email_id,role_id,dob; 

希望工程

Answer 3

您可以使用ROW_NUMBER（）的創建日期訂購併過濾掉重複的記錄

首先查詢給人紀錄minimun創作日期

;WITH cte AS (
SELECT cust_id, email_id, role_id, dob, creation_date , 
     ROW_NUMBER() OVER(PARTITION BY email_id, role_id, dob ORDER BY creation_date) seq FROM customer 
) 
SELECT cust_id, email_id, role_id, dob, creation_date 
FROM cte 
WHERE seq = 1 

對於最大創建日期相同的查詢工作與ORDER BY遞減順序進行

;WITH cte AS (
    SELECT cust_id, email_id, role_id, dob, creation_date , 
      ROW_NUMBER() OVER(PARTITION BY email_id, role_id, dob ORDER BY creation_date DESC) seq FROM customer 
    ) 
    SELECT cust_id, email_id, role_id, dob, creation_date 
    FROM cte 
    WHERE seq = 1 

編輯 對於連接查詢只需要添加加盟CTE表達SELECT語句

;WITH cte AS (
    SELECT c.cust_id, c.email_id, c.role_id, i.dob, c.creation_date , 
      ROW_NUMBER() OVER(PARTITION BY c.email_id, c.role_id, c.dob ORDER BY c.creation_date) seq 
FROM customer c 
JOIN INDIVIDUAL i on c.individaul_id=i.individaul_id 
) 
SELECT cust_id, email_id, role_id, dob, creation_date 
FROM cte 
WHERE seq = 1 

對於最大值創建日期相同的查詢條件可與ORDER BY的下降做order

;WITH cte AS (
    SELECT c.cust_id, c.email_id, c.role_id, i.dob, c.creation_date , 
      ROW_NUMBER() OVER(PARTITION BY c.email_id, c.role_id, c.dob ORDER BY c.creation_date DESC) seq 
FROM customer c 
JOIN INDIVIDUAL i on c.individaul_id=i.individaul_id 
    ) 
    SELECT cust_id, email_id, role_id, dob, creation_date 
    FROM cte 
    WHERE seq = 1 

Answer 4

您可以使用ROW_NUMBER和PARTITION。只是谷歌它相同。

入住此查詢：

Declare @customer table(cust_id int, email_id varchar(200), role_id int, dob datetime, creation_date datetime) 

    Insert into @customer 
    values(1,'abc@abc.com',5,'04-feb-1966','17-sep-2016'), 
    (2,'abc@abc.com',5,'04-feb-1966','20-sep-2016'), 
    (3,'xyz@xyz.com',5,'15-feb-1991','18-sep-2016'), 
    (4,'xyz@xyz.com',5,'15-feb-1991','21-sep-2016'), 
    (5,'pqr@pqr.com',5,'16-feb-1985','30-sep-2016'), 
    (6,'pqr@pqr.com',5,'16-feb-1985','05-nov-2016'), 
    (7,'pqr@pqr.com',5,'16-feb-1985','04-nov-2016') 

--using row number and partition to group data and remove duplicate 
    ;with custCTE as(
    select cust_id, email_id, role_id,dob,creation_date,row_number() over(partition by email_id, role_id, dob order by creation_date) as rnk 
    from @customer 
    ) 

    delete from @customer where cust_id in (select cust_id from custCTE where rnk <> 1) 
    select * from @customer 

Answer 5

這裏是解決方案。

DECLARE @MainTable TABLE 
(
    Cust_Id INT, 
    Email_Id NVARCHAR(250), 
    Role_Id INT, 
    DOB DATE, 
    Creation_Date DATE 
) 

DECLARE @Table1 TABLE 
(
    Cust_Id INT, 
    Email_Id NVARCHAR(250), 
    Role_Id INT, 
    DOB DATE, 
    Creation_Date DATE 
) 

DECLARE @Table2 TABLE 
(
    Cust_Id INT, 
    Email_Id NVARCHAR(250), 
    Role_Id INT, 
    DOB DATE, 
    Creation_Date DATE 
) 

INSERT INTO @MainTable 
     (Cust_Id , 
      Email_Id , 
      Role_Id , 
      DOB , 
      Creation_Date 
     ) 
VALUES (1 , N'abc@abc.com' , 5 , '2/4/1966' , '09/17/2016'), 
     (2 , N'abc@abc.com' , 5 , '2/4/1966' , '09/20/2016'), 
     (3 , N'xyz@xyz.com' , 5 , '2/15/1991' , '09/18/2016'), 
     (4 , N'xyz@xyz.com' , 5 , '2/15/1991' , '09/21/2016'), 
     (5 , N'pqr@pqr.com' , 5 , '2/16/1985' , '09/30/2016'), 
     (6 ,N'pqr@pqr.com' , 5 , '2/16/1985' , '11/05/2016'), 
     (7 , N'pqr@pqr.com' , 5 , '2/16/1985' , '11/04/2016') 

;WITH MainTable AS (
SELECT 
    Cust_Id , 
    Email_Id , 
    Role_Id , 
    DOB , 
    Creation_Date , 
    RANK() OVER (PARTITION BY Email_Id, Role_Id, DOB ORDER BY Creation_Date) AS [Rank] 
FROM @MainTable 
) 
INSERT INTO @Table1 
SELECT 
    MainTable.Cust_Id , 
    MainTable.Email_Id , 
    MainTable.Role_Id , 
    MainTable.DOB , 
    MainTable.Creation_Date 
FROM MainTable 
WHERE MainTable.[Rank] = 1 

;WITH MainTable AS (
SELECT 
    Cust_Id , 
    Email_Id , 
    Role_Id , 
    DOB , 
    Creation_Date , 
    RANK() OVER (PARTITION BY Email_Id, Role_Id, DOB ORDER BY Creation_Date) AS [Rank] 
FROM @MainTable 
) 
INSERT INTO @Table2 
SELECT 
    MainTable.Cust_Id , 
    MainTable.Email_Id , 
    MainTable.Role_Id , 
    MainTable.DOB , 
    MainTable.Creation_Date 
FROM MainTable 
WHERE MainTable.[Rank] <> 1 


SELECT * FROM @MainTable ORDER BY Cust_Id 
SELECT * FROM @Table1 ORDER BY Cust_Id 
SELECT * FROM @Table2 ORDER BY Cust_Id 

基本上這樣的問題可以通過使用sql server窗口函數更好地處理。窗口函數在具有添加功能的sql server 2012中更加有用。因此，上面的代碼將在MSSQL 2012中正常工作。