-

Rank vs Dense Rank:

Introduction
In this article, you will learn about RANK, DENSE_RANK and ROW_NUMBER functions in SQL Server.

Explanation

  • The RANK, DENSE_RANK and ROW_NUMBER functions are used to get the increasing integer value, based on the ordering of rows by imposing ORDER BY clause in SELECT statement.
  • When we use RANK, DENSE_RANK or ROW_NUMBER functions, the ORDER BY clause is required and PARTITION BY clause is optional.
  • When we use PARTITION BY clause, the selected data will get partitioned, and the integer value is reset to 1 when the partition changes.
Let’s see an example.
For this example, I have created a table called Employees in which I have added three columns EMPID, Name, and Salary respectively. To create table and column, the script is as given below.

Creating a table 

  1. CREATE TABLE [dbo].[Employee](  
  2.     [EMPID] [nvarchar](30) NOT NULL,  
  3.     [Name] [nvarchar](150) NULL,  
  4.     [Salary] [money] NULL  
  5. )   

Insert some data into this table, script as follows

  1. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP101', 'Vishal', 15000)  
  2. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP102', 'Sam', 20000)  
  3. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP105', 'Ravi', 10000)  
  4. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP106', 'Mahesh', 18000)  
The table is created and data is inserted. Now, let’s see how these about RANK, DENSE_RANK and ROW_NUMBER functions behave. For this, we can write Select statement as shown below.
  1. select EMPID, Name,Salary,  
  2. RANK() over (order by Salary desc) as _Rank,  
  3. DENSE_RANK () over (order by Salary desc) as DenseRank ,  
  4. ROW_NUMBER() over (order by Salary desc) as RowNumber from Employee  
You can see the below output that RANK, DENSE_RANK and ROW_NUMBER functions give the same output and returning integer values based on highest salary.

SQL
I am assuming you have got what RANK, DENSE_RANK and ROW_NUMBER functions does.
Now, let’s see what is the difference between those functions. For this, we need to insert some duplicate salary into this table. The script below will insert some more data with same salary.
  1. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP108', 'Rahul', 20000)  
  2. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP109', 'menaka', 15000)  
  3. INSERT [dbo].[Employee] ([EMPID], [Name], [Salary]) VALUES ('EMP111', 'akshay', 20000)  
Now, you can see the table which has got some duplicate salary records.

SQL
Now, we can see the different between RANK, DENSE_RANK and ROW_NUMBER functions by executing the same SELECT statement which we executed before.

SQL
Here, in above figure, you can see that ROW_NUMBER function does not have any changes. It keeps increasing integer by one and it is not caring about duplicate values.
In RANK, DENSE_RANK function, it is looking for duplicate values. The integer value is increasing by one but if the same value (Salary) is present in the table, then the same integer value is given to all the rows having the same value(Salary), as marked in sky blue color.
In RANK function, the next row after the duplicate values (salary),marked in red color, will not give the integer value as next rank but instead of it, it skips those ranks and gives what is the next incremented rank. In the above case, the first three values are having same salary so it gives same rank to them but in next row, it gives as 4, It skips two and three as first three rows have same ranks.
In DENSE_RANK function, it will not skip any rank. This means the next row after the duplicate value (salary) rows will have the next rank in the sequence.

Some of the use of these functions in real-time

  • Using ROW_NUMBER function, we can delete duplicate data from table. For this, we need to insert some duplicate records into this table.

    SQL
In order to remove all duplicate rows, we need to keep one row and remaining rows we need to delete. For this, we are using ROW_NUMBER function as it will be easy.
  1. with empCTE as  
  2. (  
  3. select *, ROW_NUMBER() over(partition by EMPID order by EMPID) as rowno from Employee  
  4. )  
  5. delete from empCTE where rowno>1  
In this code, I have written Select statement to get a rank and I made as a CTE. After that, I am deleting records from CTE where row number is greater than one so that one record remains.

SQL
As you can see in the above figure, the duplicate rows are deleted.
  • Using RANK or DENSE_RANK function, we can find Nth highest salary very easily, however which function to use depends on what we need.
For this, let’s update Ravi’s salary same as Sam. The script is as follows
  1. update Employee set Salary=20000 where EMPID='EMP105'  
Now, we have two first height salary records. Let’s write a Select statement to find second highest salary using RANK or DENSE_RANK function as shown below.
SQL
As you can see above, the DENSE_RANK function is giving an accurate result and RANK function is not returning any row. So, we can use DENSE_RANK Function even if there is a same salary .

Comments

Post a Comment

Popular posts from this blog

-
Image
CLUSTERED INDEX VS NON CLUSTERED INDEX Clustered Index Only one clustered index can be there in a table because of the sorted order Sort the records and store them physically according to the order Data retrieval is faster than non-clustered indexes Do not need extra space to store logical structure A primary key by default creates a clustered index on it. Non Clustered Index There can be any number of non-clustered indexes in a table because of the number of unique keys we can create. Do not affect the physical order. Create a logical order for data rows and use pointers to physical data files Data insertion/update is faster than clustered index Use extra space to store logical structure A unique key/keys by dafault created an unclustered index on it. In Depth Explanation: Clustered Technically speaking, a clustered index is a B-Tree data structure where all the rows in a table are stored at the leaf level of the index. In other words, a clustered ind...
Read more
-
Image
SQL JOINS Tutorial: INNER, OUTER, LEFT, RIGHT, CROSS What are JOINS? Joins help retrieving data from two or more database tables. The tables are mutually related using primary and foreign keys. id first_name last_name movie_id 1 Adam Smith 1 2 Ravi Kumar 2 3 Susan Davidson 5 4 Jenny Adrianna 8 6 Lee Pong 10 id title category 1 ASSASSIN'S CREED: EMBERS Animations 2 Real Steel(2012) Animations 3 Alvin and the Chipmunks Animations 4 The Adventures of Tin Tin Animations 5 Safe (2012) Action 6 Safe House(2012) Action 7 GIA 18+ 8 Deadline 2009 18+ 9 The Dirty Picture 18+ 10 Marley and me Romance   Types of joins Cross JOIN Cross JOIN is a simplest form of JOINs which matches each row from one database table to all rows of another. In other words it gives us combinations of each row of first table with all records in second table. Suppose we want to get all member records against all the movie records, we can use the script shown below to ge...
Read more
-
Image
COMMON TABLE EXPRESSIONS(cte): It is used best used to extract the hierarchical data  and it is best explained with an example: Problem Many organizations have some type of hierarchy for business processes. When it comes to large organizations, the hierarchy can get very complex and large, so building a hierarchy in a RDBMS is a tedious task. We have to create views, cursors and so on, but using a CTE in SQL Server is a better solution to retrieve hierarchy-based data. Solution Common Table Expressions (CTE) have two types, recursive and non-recursive. We will see how the recursive CTE works with examples. A recursive CTE can be explained in three parts: Anchor Query: This is the first statement which is executed. This query will give the base data for the CTE. Separator: This is the middle part where in we generally use a UNION ALL and few more operators. Recursive Query: This is the main part, this is the CTE query which refers to the same CTE by recursion. ...
Read more