Ranking in T-SQL

25 Sep 2012

In Queries, we need to rank and number records. SQL Server gives you a few functions to rank your records.

RANK

Rank we can use to rank data in a normal way. In this ranking, if we have two 1 next to one will have rank 3.

Name	Marks	Rank
John	75	1
Mark	75	1
Steve	64	3
Gates	54	4

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('John',75),('Mark',75),('Steve',64),('Gates',54)
GO

--Ranking data over Score
SELECT *, RANK() OVER (ORDER BY [Score] DESC) AS [RANK] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

This ranks based on the column that you specify by OVER.

DENSE_RANK

Dense Rank is not missing ranks like in normal rank function. It has continuity over ranking.

Name	Marks	Dense Rank
John	75	1
Mark	75	1
Steve	64	2
Gates	54	3

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('John',75),('Mark',75),('Steve',64),('Gates',54)
GO

--Ranking data over Score
SELECT *, DENSE_RANK() OVER (ORDER BY [Score] DESC) AS [DENSE RANK] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

ROW NUMBER

Row Number don’t worry much about ranking it just give an incremental number to row. It’s simple.

Name	Marks	Row Number
John	75	1
Mark	75	2
Steve	64	3
Gates	54	4

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('John',75),('Mark',75),('Steve',64),('Gates',54)
GO

--Numbering data over Score
SELECT *, ROW_NUMBER() OVER (ORDER BY [Score] DESC) AS [ROW NUMBER] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

NTile

NTile will separate your code into the number that you parse. If you want to group your results into 3 groups you can use NTILE(3). If you want to divide your results into 7 you can use NTILE(7).

Name	Mark	NTile(3)
Dedunu	89	1
Dhananjaya	85	1
Hasitha	75	2
Shamil	74	2
Sanjana	65	3
Anuradha	55	3

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('Dedunu',89),('Dhananjaya',85),('Hasitha',75),('Shamil',74),('Sanjana',65),('Anuradha',55)
GO

--NTile(3) over Score
SELECT *, NTILE(3) OVER (ORDER BY [Score] DESC) AS [NTile(3)] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

If you are going to group them into 4, you will have this result. It always tries to round to the lower numbers.

Name	Mark	NTile(4)
Dedunu	89	1
Dhananjaya	85	1
Hasitha	75	2
Shamil	74	2
Sanjana	65	3
Anuradha	55	4

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('Dedunu',89),('Dhananjaya',85),('Hasitha',75),('Shamil',74),('Sanjana',65),('Anuradha',55)
GO

--NTile(4) over Score
SELECT *, NTILE(4) OVER (ORDER BY [Score] DESC) AS [NTile(4)] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

Then if you are going to group odd numbered records by even or even numbered records by odd, first tile will have an extra record.

Name	Mark	NTile(5)
Dedunu	89	1
Dhananjaya	85	1
Hasitha	75	2
Shamil	74	3
Sanjana	65	4
Anuradha	55	5

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('Dedunu',89),('Dhananjaya',85),('Hasitha',75),('Shamil',74),('Sanjana',65),('Anuradha',55)
GO

--NTile(5) over Score
SELECT *, NTILE(5) OVER (ORDER BY [Score] DESC) AS [NTile(5)] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

If you are trying to group records by more than records number or same number it will act like ROW_NUMBER function.

Name	Mark	NTile(>=6)
Dedunu	89	1
Dhananjaya	85	2
Hasitha	75	3
Shamil	74	4
Sanjana	65	5
Anuradha	55	6

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('Dedunu',89),('Dhananjaya',85),('Hasitha',75),('Shamil',74),('Sanjana',65),('Anuradha',55)
GO

--NTile(6) data over Score
SELECT *, NTILE(6) OVER (ORDER BY [Score] DESC) AS [NTile(6)] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

USE [tempdb]
GO
--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Score]) 
    VALUES ('Dedunu',89),('Dhananjaya',85),('Hasitha',75),('Shamil',74),('Sanjana',65),('Anuradha',55)
GO

--NTile(8) over Score
SELECT *, NTILE(8) OVER (ORDER BY [Score] DESC) AS [NTile(8)] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

PARTITION BY

Let’s move to PARTITION BY clause. If you want to categorize your ranks into any other factor, you can use PARTITION BY clause. It lets you have separate rankings on a factor that you want.

Name	Subject	Mark	Rank
Hasitha	Eng	75	1
Anuradha	Eng	55	2
Dedunu	Math	89	1
Dhananjaya	Math	85	2
Sanjana	Math	65	3
Shamil	Sci	74	1

USE [tempdb]
GO

--Creating table Marks
CREATE TABLE [MARKS](
    [Name] varchar(25),
    [Subject] varchar(25),
    [Score] int)
GO

--Inserting values into Mark table
INSERT INTO [MARKS]([Name], [Subject], [Score]) 
    VALUES ('Dedunu','Math',89),('Dhananjaya','Math',85),('Hasitha','Eng',75),('Shamil','Sci',74),('Sanjana','Math',65),('Anuradha','Eng',55)
GO

--Partitioning rank by Suject, and ranking over Score
SELECT *, RANK() OVER (PARTITION BY [Subject] ORDER BY [Score] DESC) AS [RANK] FROM MARKS

--Dropping table
DROP TABLE [MARKS]
GO

Ranking in T-SQL

Tags