In this post, I am comparing and sharing maths on ROW_NUMBER(), RANK(), and DENSE_RANK() of SQL Server. We are using these functions in day to day work, but still few people are not aware of it.

Something like, Can we use DENSE_RANK() as ROW_NUMBER()?
The answer is yes, we can use for DISTINCT records because using ROW_NUMBER(), you cannot generate distinct records as it applicable for all records.

Check the below samples, and try it yourself:

Create a table with sample records:

CREATE TABLE tbl_numbers (name varchar(10))
GO

INSERT INTO tbl_numbers 
VALUES ('ABC'),('ABC'),('ABC')
,('XYZ'),('XYZ'),('LMN'),('OPQ'),('OPQ')
GO

Compare the output of Row_Number, Rank, DENSE_RANK:

SELECT 
	name
	,ROW_NUMBER() OVER(ORDER BY name) AS RowNumber
	,RANK() OVER(ORDER BY name) AS RankNumber
	,DENSE_RANK() OVER(ORDER BY name) AS DENSE_RankNumber
FROM tbl_numbers

Result:

name       RowNumber            RankNumber           DENSE_RankNumber
---------- -------------------- -------------------- --------------------
ABC        1                    1                    1
ABC        2                    1                    1
ABC        3                    1                    1
LMN        4                    4                    2
OPQ        5                    5                    3
OPQ        6                    5                    3
XYZ        7                    7                    4
XYZ        8                    7                    4

Using ROW_NUMBER, you cannot get the distinct records:

SELECT DISTINCT
	name
	,ROW_NUMBER() OVER(ORDER BY name) AS RowNumber
FROM tbl_numbers

Result:

name       RowNumber
---------- ------------
ABC        1
ABC        2
ABC        3
LMN        4
OPQ        5
OPQ        6
XYZ        7
XYZ        8

Using CTE, get the distinct records:

WITH CTE AS
(
SELECT DISTINCT
	name
FROM tbl_numbers
)
SELECT 
	name
	,ROW_NUMBER() OVER(ORDER BY name) AS RowNumber
FROM CTE

Result:

name       RowNumber
---------- -----------
ABC        1
LMN        2
OPQ        3
XYZ        4

Using DENSE_RANK, get the distinct records:
This query is faster than, CTE and you can use DENSE_RANK as a ROW_NUMBER for distinct Records.

SELECT DISTINCT
	name
	,DENSE_RANK() OVER(ORDER BY name) AS DENSE_RankNumber
FROM tbl_numbers

Result:

name       DENSE_RankNumber
---------- --------------------
ABC        1
LMN        2
OPQ        3
XYZ        4

Are row_number and partition by available in MySQL?

Answer is NO.

Microsoft SQL Server support row_number and partition by for finding the unique record from a set of duplicate records. We can also set the order of data.

MySQL doesn’t provide this feature directly. We can achieve this by using an inline variable for SELECT statements.

Below is a full demonstration:

I prepared a few scripts for sample data and stored procedure for finding the only unique records.

Below is a full demonstration:

/*Create sample database and table with sample data.*/
CREATE DATABASE `Employee`;

CREATE TABLE Employee.EmpDuplicate
(
	EmpRemarkID INTEGER
	,EmpID INTEGER
    ,EmpRemark VARCHAR(50)
);

INSERT INTO Employee.EmpDuplicate VALUES
(1,25,'abc'),(2,25,'abc'),(3,25,'abc')
,(4,26,'xyz'),(5,27,'pqr'),(6,27,'pqr')
,(7,27,'pqr'),(8,27,'pqr'),(9,28,'xvk')
,(10,28,'xvk'),(11,29,'pqr'),(12,29,'pqr')
,(13,29,'abc'),(14,30,'abc');

/*Sample Stored Procedure to find Unique record from the group of duplicate.*/
DELIMITER $$
CREATE PROCEDURE Employee.usp_GETUniqueEmployeeRecords()
BEGIN

    SET @rn1:=1;
    SET @pk1:='';    				
    SELECT  
        EmpRemarkID
	,EmpID
	,EmpRemark				
    FROM
    (
	SELECT  
	   EmpRemarkID
	   ,EmpID
	   ,EmpRemark
	   ,@rn1 := IF(@pk1=CONCAT(EmpID,EmpRemark), @rn1+1,1) AS rowNumber
	   ,@pk1 := CONCAT(EmpID,EmpRemark)         
	FROM
	(
	    SELECT  
         	EmpRemarkID
		,EmpID
		,EmpRemark
	    FROM Employee.EmpDuplicate
	) A
    ) B WHERE rowNumber=1;			
END$$
DELIMITER ;

/*Execute this Stored Procedure.*/
CALL Employee.usp_GETUniqueEmployeeRecords()

Results:

In this post, I am sharing a script to copy a unique record from one table to another.

You can perform this activity using ROW_NUMBER() and PARTITION BY clause.

This partition clause identified set of duplicate records and row_number assign an ID for this set.

Below is a full demonstration:

Let’s first create table with some records:

CREATE TABLE dbo.tbl_First
(
 ID INTEGER IDENTITY(1,1)
 ,Name VARCHAR(250)
)
GO
INSERT INTO dbo.tbl_First(Name) VALUES('ABC'),('XYZ'),('ABC'),('EFG'),('PQR'),('EFG'),('RND'),('YFQ'),('XVK'),('ABC')
GO
SELECT *FROM dbo.tbl_First ORDER BY Name 
GO

Result of above select statement is:

You can see ABC and EFG are duplicate name.

Create a second table with copy unique data from the first table:

CREATE TABLE dbo.tbl_Second
(
 ID INTEGER 
 ,Name VARCHAR(250)
)
GO
 
INSERT INTO dbo.tbl_Second
SELECT 
 ID
 ,Name   
FROM
(
 SELECT 
  ID
  ,Name
  ,ROW_NUMBER() OVER (PARTITION BY Name ORDER BY ID) AS RowNumber
 FROM  dbo.tbl_First AS T
)AS T WHERE T.RowNumber = 1 
GO

SELECT *FROM dbo.tbl_Second
GO

The result of above second table is:

Now you can see all names are unique and copied only seven unique records.