Window Functions in SQL

Learning Objectives

By the end of this lesson, you will be able to:

Explain window functions and various clauses

List the aggregate window functions

Classify ranking window functions

Categorize the miscellaneous window functions

Introduction to Window Functions
 Window Functions

The window functions is like an SQL function that takes input values from a window of one or
more rows of a SELECT statement’s result set.

The window functions perform various operations on a group of rows and provide an
aggregated value for each row with a unique identity.
 General Syntax

Syntax
window_function_name(expression)OVER([partition_definition]
[order_definition] [frame_definition] )

• Window function specifies the window function name

with an expression.

• OVER clause can comprise partition definition, order definition, and

frame definition.
 Types of Clauses

Partition
Clause

Clauses
Frame Order By
Clause Clause
 Partition Clause

The partition clause is used to divide or split the rows into partitions, and
the partition boundary is used to split two partitions.

Syntax
PARTITION BY
<expression>[{,<expression>...}]
 Order By Clause

Order By clause is an arrangement of rows inside a partition. It performs

partitions using multiple keys where each key has an expression.

Syntax
ORDER BY <expression> [ASC|DESC],
[{,<expression>...}]
 Frame Clause

Frame clause is defined as subset of the current position. It allows to move the
subset within a partition based on the position of the current row in its partition.

Syntax

frame_unit

{<frame_start>|<frame_between>}

The frame unit can be a row or range that specifies the kind of relationship
between the current row and frame row.
 Frame Clause

Keywords Meaning

Frame unit Rows: It assigns row number for offset of current and frame row.

Frame unit Range: It assigns row values for offset of current and frame row.

Frame start Frame start specifies the frame boundary.

Frame between Frame between specifies the frame boundary.

 Use Case for Window Functions

Problem Scenario:

The HR of a company wants to calculate the performance of employees department-wise based

on the employee ratings.

Objective:

You are required to retrieve the employee ID, first name, role, department, and employee rating
by calculating the maximum employee rating using PARTITION BY and MAX function on
department and employee rating fields respectively.

Instructions:

Refer to the employee dataset given in the course resource section in LMS and create an
employee table using fields mentioned in dataset. Insert the values accordingly to perform the
above objectives.
 Use Case for Window Functions

Field Name Description

EMP_ID Employee ID

FIRST_NAME First name of the employee

LAST_NAME Last name of the employee

GENDER Gender of the employee (M/F)

Designation of the employee (Junior, Senior, Lead, and Associate Data

ROLE Scientist)

DEPT Name of the department (Retail, Finance, Automotive, and Healthcare)

 Use Case for Window Functions

Field Name Description

EXP Experience of the employee

COUNTRY Country where the employee lives

CONTINENT Continent based on the country

SALARY Salary of the employee per month in dollars

Rating for the employee (1: Not achieved any goals, 2: Below
EMP_RATING expectation, 3: Meeting expectation, 4: Excellent performance,
5: Overachiever)

MANAGER_ID Employee ID for the manager

 Use Case for Window Functions

Solution:

# SELECT EMP_ID, FIRST_NAME,ROLE,DEPT,EMP_RATING and calculate the maximum EMP_RATING

in a department from the employee table using partition clause on department, Max
function.

SELECT EMP_ID,FIRST_NAME,ROLE,DEPT,EMP_RATING,MAX(EMP_RATING)OVER(PARTITION BY DEPT) AS

MAX_EMP_RATING FROM emp_table;

By executing this query, the HR can identify the maximum rating of the
employee in a department.
 Use Case for Window Functions

Output:
Aggregate Window Functions
 Aggregate Window Functions

The aggregate window functions perform on a particular set of rows and

provide the result in a single row.

Syntax
window_function ( [ ALL ] expression )
OVER ( [ PARTITION BY expr_list ] [
ORDER BY order_list frame_clause ] )
 Arguments in Aggregate Window Functions

Keywords Meaning

Window function It can be any aggregate window function.

ALL ALL helps to maintain all duplicate values from the expression.

OVER It distinguishes window aggregation from general aggregation functions.

PARTITION BY PARTITION BY provides a window if there are one or more expressions.

ORDER BY ORDER BY is used to sort the rows within each partition.

 Types of Aggregate Window Functions

AVG()

01 MIN()
02

COUNT() 05
03 MAX()
0
4 04

SUM()
 Types of Aggregate Window Functions

Window
Argument Type Return Type Description
Function

BINARY, DECIMAL, Same as It returns the

VARCHAR, DATE, argument type minimum value of
MIN()
TIME, or the expression
TIMESTAMP across all input
values.

BINARY, Same as It returns the

DECIMAL, argument type maximum value of
MAX() VARCHAR, DATE, the expression
TIME, or across all input
TIMESTAMP values.
 Types of Aggregate Window Functions

Window Argument Type Return Type Description

Function

SMALLINT,
DECIMAL argument: It returns average
INTEGER, BIGINT,
AVG() returns in decimal value for the input
FLOAT, DOUBLE,
Other types: double expression values.
DECIMAL,
INTERVALYEAR or
INTERVALDAY
It counts the
All argument data BIGINT number of input
COUNT()
types rows.
 Types of Aggregate Window Functions

Window
Function Argument Type Return Type Description

SMALLINT, DECIMAL It returns the

INTEGER, BIGINT, argument: returns sum of the
SUM() FLOAT, DOUBLE, in decimal expression
DECIMAL, Float: returns in across all input
INTERVALDAY, or double values.
INTERVALYEAR Other types :
BIGINT
 Use Case for MIN and MAX

Problem Scenario:

The HR of a company wants to identify the minimum and the maximum salary of the employees in
a role.

Objective:

You are required to display the employee’s ID, first name, role, and salary by finding the minimum
and maximum salary of the employees using PARTITION BY clause, MIN, and MAX functions on
role and salary fields respectively.

Instructions:

Refer to the employee table which is created and perform the above objectives.
 Use Case for MIN and MAX

Solution:

# SELECT EMP_ID, FIRST_NAME, ROLE, SALARY and calculate minimum, maximum salary of the
employees using PARTITION CLAUSE on the role field, MIN , MAX function.

SELECT EMP_ID,FIRST_NAME,ROLE,SALARY, MAX(SALARY) OVER (PARTITION BY ROLE)

MAX_SALARY,MIN(SALARY) OVER (PARTITION BY ROLE) MIN_SALARY FROM emp_table

By executing this query, the HR can identify the maximum and the minimum salary in a role.
 Use Case for MIN and MAX

Output:
 Use Case for AVG and COUNT

Problem Scenario:

The HR of a company wants to identify the average performance of the employee's department-
wise and also find the total number of records in a department.

Objective:

You are required to display the employee’s ID, first name, department, and employee rating by
calculating the average employee rating and the total number of records in a department using
PARTITION BY clause, AVG, and COUNT functions on department and employee rating fields
respectively.

Instructions:

Refer to the employee table which is created and perform the above objectives.
 Use Case for AVG and COUNT

Solution

# SELECT EMP_ID, FIRST_NAME,DEPT,EMP_RATING and calculate the average employee rating

and total no of records in a department using AVG,COUNT function.

SELECT EMP_ID,FIRST_NAME,DEPT,EMP_RATING, AVG(EMP_RATING)OVER(PARTITION BY DEPT)

AVG_EMP_RATING_IN_DEPT, COUNT(*) OVER(PARTITION BY DEPT) NO_OF_RECORDS_IN_DEPT FROM
emp_table

By executing this query, the HR can identify the average performance of the department
and the total number of records in a department.
 Use Case for AVG and COUNT

Output:
 Use Case for SUM

Problem Scenario:

The HR of a company wants to calculate the total employee rating in a department.

Objective:

You are required to display the employee’s Id, first name, department, and employee rating by
calculating the total employee rating in a department using PARTITION BY clause and SUM
function on the department and the employee rating fields respectively.

Instructions:

Refer to the employee table which is created and perform the above objectives.
 Use Case for SUM

Solution:

# SELECT EMP_ID, FIRST_NAME, DEPT,EMP RATING and calculate the total employee rating in a
department using PARTITION CLAUSE on a dept and SUM function.

SELECT EMP_ID,FIRST_NAME,DEPT,EMP_RATING, SUM(EMP_RATING) OVER (PARTITION BY DEPT)

TOTAL_EMP_RATING_IN_DEPT FROM emp_table

By executing this query, the HR can identify the total employee rating in a department .
 Use Case for SUM

Output:
Assisted Practice: Aggregate Window Functions

Duration: 20 min
Problem Statement: You are required to calculate the total, average, maximum, and minimum salary of the
employee by grouping the departments from the employee table.
Assisted Practice: Aggregate Window Functions

Steps to be performed:
Step 1: Creating the employee table and inserting values in it:

CREATE
CREATE TABLE lep_7.employee ( emp_id int NOT NULL, f_name varchar(45) NULL,
l_name varchar(45) NOT NULL, job_id varchar(45) NOT NULL, salary
decimal(8,2) NOT NULL, manager_id int NOT NULL, dept_id varchar(45) NOT
NULL, PRIMARY KEY(emp_id));

INSERT
INSERT INTO lep_7. employee
(emp_id,f_name,l_name,job_id,salary,manager_id,dept_id) VALUES
('103','krishna','gee','125','500000','05','44');
Assisted Practice: Aggregate Window Functions

Step 2: Querying to perform aggregate window functions:

QUERY

SELECT dept_id,salary,SUM(salary) OVER (PARTITION BY dept_id) salary_total,

AVG(salary) OVER(PARTITION BY dept_id) score_avg , MAX(salary) OVER
(PARTITION BY dept_id) max_salary , MIN(salary) OVER (PARTITION BY dept_id)
min_salary FROM employee
Assisted Practice: Aggregate Window Functions

Output:
Ranking Window Functions
 Ranking Window Functions and Its Types

Ranking window functions specify the rank for individual fields as per the
categorization.

Dense rank Rank

Ranking
window
functions

Row number Percent rank

 Dense Rank

Definition Syntax

DENSE_RANK() OVER (
• It assigns a rank to every row in a
partition based on the ORDER BY PARTITION BY
clause. <expression>[{,<expression>...}]

ORDER BY <expression>
• It assigns the same rank for equal [ASC|DESC], [{,<expression>...}])
values.

• It has no gaps if two or more

rows have a similar rank.
 Rank

Definition Syntax

• Rank helps to assign a rank to RANK() OVER (

PARTITION BY
all rows within every partition. <expr1>[{,<expr2>...}]
ORDER BY <expr1>
[ASC|DESC], [{,<expr2>...}]
• The first row of the rank will )
be 1.

• Same rank for the same value.

• There will be a gap if two or

more rows have the same
rank.
 Use Case for Rank and Dense Rank

Problem Scenario:

The HR of a company wants to assign a rank for each employee based on their employee rating.

Objective:

You are required to display the employee’s ID, first name, department, and employee rating by
assigning a rank to all the employees based on their employee rating using ORDER BY clause,
RANK, and DENSE RANK functions on the employee rating field.

Instructions:

Refer to the employee table which is created and perform the above objectives.
 Use Case for Rank and Dense Rank

Solution:

# SELECT EMP_ID, FIRST_NAME, DEPT,EMP RATING and assign a rank to all the employee
based on their employee rating using Rank and Dense Rank .

SELECT EMP_ID,FIRST_NAME,DEPT,EMP_RATING, RANK() OVER (ORDER BY EMP_RATING)

EMP_RATING_RANK, DENSE_RANK() OVER (ORDER BY EMP_RATING) EMP_RATING_DENSE_RANK FROM
emp_table;

By executing this query, the HR can identify the rank of the employees based on their
employee rating.
 Use Case for Rank and Dense Rank

Output:
 Row Number

Definition Syntax

• Row number retrieves the ROW_NUMBER() OVER

(<partition_definition>
unique sequential number of <order_definition>)
each row for the specified
data.
• Similar values will have
different ranks.
 Use Case for Row Number

Problem Scenario:

The IT department of a company wants to assign an asset number for each employee based on
their employee ID in ascending order.

Objective:

You are required to display the employee’s ID, first name, role, and department by assigning a
number to each employee in ascending order of their employee ID using ORDER BY clause and
ROW NUMBER function on the employee ID field.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for Row Number

Solution:

# SELECT EMP_ID, FIRST_NAME, ROLE, DEPT and assign assetnumber to all the employee in
ascending order of their employee ID.

SELECT EMP_ID,FIRST_NAME,ROLE,DEPT,ROW_NUMBER() OVER(ORDER BY EMP_ID)

EMP_ID_ASC_ROWNUMBER FROM emp_table;

By executing this query, the IT department can identify a asset number assigned to each
employee in ascending order of their employee ID.
 Use Case for Row Number

Output:
 Percent Rank

Definition Syntax

• Percent rank helps to PERCENT_RANK()OVER (

PARTITION BY expr,...
evaluate the percentile rank ORDER BY expr
[ASC|DESC],...)
of a value in a partition or
result set.

• It returns a value between

zero to one.
 Use Case for Percent Rank

Problem Scenario:

The HR of a company wants to calculate the overall percentile of the employee rating in a
department.

Objective:

You are required to display employee’s ID, first name, role, department, and employee rating by
calculating the percentile of the employee rating in a department using ORDER BY clause and
PERCENT RANK function on an employee rating field.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for Percent Rank

Solution:

# SELECT EMP_ID,FIRST_NAME,ROLE,DEPT,EMP_RATING and calculate the percentile of the

employee rating using ODERBY and PERCENT RANK function.

SELECT EMP_ID,FIRST_NAME,ROLE,DEPT,EMP_RATING,PERCENT_RANK() OVER(ORDER BY EMP_RATING)

PERCENTILE_EMP_RATING FROM emp_table;

By executing this query, the HR can identify the percentile score of each employee in a
department.
 Use Case for Percent Rank

Output:
Miscellaneous Window Functions
 Types of Miscellaneous Window Functions

FIRST VALUE ()

LAST VALUE ()
NTILE ()

LEAD () NTH VALUE ()

LAG () CUME DIST ()

 First Value Function

Definition Syntax

FIRST_VALUE (expression) OVER

First value function returns the ( [partition_clause]
value of the expression from the [order_clause] [frame_clause]
first row of the window frame. )
 Use Case for First Value Function

Problem Scenario:

The HR department of an organization aims to find the employee ID of the employee with the
highest experience by sorting their experience in descending order.

Objective:

You are required to display the employee ID, first name, and experience, as well as identify the
employee ID of the first employee by sorting the experience in descending order using the
ORDER BY clause and first value function on the experience and employee ID fields respectively.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for First Value Function

Solution:

# SELECT EMP_ID,FIRST_NAME,EXP and determine the highest experience in the EMP_ID based
on descending order of the experience.

SELECT EMP_ID, FIRST_NAME, EXP, FIRST_VALUE(EMP_ID) OVER (ORDER BY EXP DESC)

highest_exp_of_emp_id FROM emp_table;

By executing this query, HR can identify the employee ID with the

highest experience.
 Use Case for First Value Function

Output:
 NTH Value Function

Definition Syntax

NTH_VALUE(expression, N)
The NTH value function acquires FROM FIRST
OVER (
a value from the Nth row of an partition_clause
ordered group of rows. order_clause
frame_clause
)
 Use Case for NTH Value Function

Problem Scenario:

The HR of a company wants to identify the third-highest experience among employees in the
company.

Objective:

You are required to display the employee’s ID, first name, and experience by calculating the third-
highest experience among employees using ORDER BY clause and NTH value function in
descending order of experience field.

Instructions:

Refer to the employee table which is created and perform the following objective.
 Use Case for NTH Value Function

Solution:

# SELECT EMP_ID,FIRST_NAME,EXP and calculate the third highest experience among

employees using order by on experience in descending order and N_TH VALUE function

SELECT EMP_ID,FIRST_NAME,EXP, NTH_VALUE(EXP,3) OVER(ORDER BY EXP DESC)

THIRD_HIGHEST_EXPERIENCE FROM emp_table

By executing this query, the HR can identify the third highest experience of the employee.
 Use Case for NTH Value Function

Output:
 NTILE Function

Definition Syntax
NTILE(n) OVER (
NTILE(n) OVER (
NTILE function breaks the rows PARTITION BY
PARTITION<expression>[{,<expression>..
BY
into a sorted partition in a <expression>[{,<expression
.}]
certain number of groups. >...}] ORDER BY <expression>
ORDER BY[ASC|DESC],
<expression>
[ASC|DESC],
[{,<expression>...}]
[{,<expression>...}]
)
)

N_TH Value

The nth value function

helps you to acquire a
value from the Nth row of
a ordered group of rows.
 Use Case for NTILE Function

Problem Scenario:

The HR of a company wants to sort the employee table in ascending based on their
experience in four partitions.

Objective:

You are required to display all the details by sorting the experience into four partitions in
ascending order using the ORDER BY clause and NTILE function on an experience field.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for NTILE Function

Solution:

# SELECT all the details in the employee table by sorting in ascending order of the
experience into four partitions using ORDER BY on EXP and NTILE function.

SELECT * , NTILE(4)OVER(ORDER BY EXP) PARTITION_BY_EXP FROM emp_table

By executing this query, the HR can view the experience in ascending order into four partitions.
 Use Case for NTILE Function

Output:
 Cume Dist Function

Definition Syntax

CUME_DIST( ) OVER ( [
The Cume Dist function partition_by_clause ]
calculates the cumulative order_by_clause )
distribution of a number in a
group of values.

N Title

Function breaks the

rows in a sorted
partition into a certain
number of groups.
 Use Case for Cume Dist Function

Problem Scenario:

The HR of a company wants to sort the employee data based on their experience in ascending
order and calculate the cumulative distribution on the employee table.

Objective:

You are required to display the employee’s ID, first name, and experience by calculating the
cumulative distribution of the experience with the help of ROW NUMBER using ORDER BY, ROW
NUMBER, and CUME DIST function on an experience field.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for Cume Dist Function

Solution:

# SELECT EMP_ID,FIRST_NAME,EXP and calculate the cumulative distribution of the

experience with the help of Row number using ODERBY and ROW_NUMBER, CUME_DIST function.

SELECT EMP_ID,FIRST_NAME,EXP, ROW_NUMBER() OVER (ORDER BY EXP) ROW_NUMBER_EXP,

CUME_DIST() OVER (ORDER BY EXP) CUME_DIST_EXP FROM emp_table;

By executing this query, the HR can identify the cumulative distribution of the experience with the help
of row number .
 Use Case for Cume Dist Function

Output:
 Lead Function

Definition Syntax

LEAD OVER() ( PARTITION BY

Lead function is used to retrieve (expr) ORDER BY (expr) )

the values from the next N

rows.
 Lag Function

Definition Syntax

LAG () OVER ( PARTITION BY

• Lag function is used to retrieve expr,... ORDER BY expr
[ASC|DESC],... )
the values from previous N
rows.

• It is the reverse of lead function.

 Use Case for Lead and Lag Function

Problem Scenario:

The HR of a company wants to ignore the two lowest and highest experiences of the employees.

Objective:

You are required to display the employee’s ID, first name, experience and sort the employees in
ascending order of their experience. Ignore the two lowest experiences using LEAD and two
highest experiences using LAG to determine the median of the employee experience.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for Lead and Lag Function

Solution:

# SELECT EMP_ID,FIRST_NAME, EXP and ignore the two lowest and highest experience using
LEAD and LAG function

SELECT EMP_ID,FIRST_NAME,EXP, LEAD(EXP,2) OVER (ORDER BY EXP) LEAD_2_LOWEST_EXP,

LAG(EXP,2) OVER (ORDER BY EXP) LAG_2_HIGHEST_EXPFROM emp_table

By executing this query, the HR can identify the median experience of the employee.
 Use Case for Lead and Lag Function

Output:
 Last Value Function

Definition Syntax

LAST_VALUE (expression) OVER

Last value function returns the last ( [partition_clause]
value of a specific column in an [order_clause] [frame_clause]
ordered sequence. )
 Use Case for Last Value Function

Problem Scenario:

The HR of a company wants to determine the last employee ID by sorting the experience in
ascending order.

Objective:

You are required to display the employee’s ID, first name, and experience and determine the last
employee ID by sorting the experience in ascending order using ORDER BY clause and last value
function on the experience and employee ID field respectively.

Instructions:

Refer to the employee table which is created and perform the above objective.
 Use Case for Last Value Function

Solution:

# SELECT EMP_ID,FIRST_NAME,EXP and determine the last value in the EMP_ID based on
ascending order of the experience.

SELECT EMP_ID,FIRST_NAME,EXP, LAST_VALUE(EMP_ID) OVER (ORDER BY EXP RANGE

BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING ) LastValue
FROM emp_table

By executing this query, the HR can identify the last value of the employee ID based on
their experience.
 Use Case for Last Value Function

Output:
Assisted Practice: Ranking and Miscellaneous Window
Functions

Duration: 15 min
Problem Statement: You are required to identify the rank and row number and calculate the cumulative
distribution and percentile score based on the student score from the marksheet table.
Assisted Practice: Ranking and Miscellaneous Window
Functions

Steps to be performed:
Step 1: Creating the marksheet table and inserting values in it:

CREATE

CREATE TABLE marksheet ( score INT NOT NULL, year INT NULL, class
varchar(45) NULL, ranking varchar(45) NULL, s_id INT NOT NULL );

- INSERT

INSERT INTO marksheet (score,year,class,ranking,s_id) VALUES

('989','2014','10','1','1');
Assisted Practice: Ranking and Miscellaneous Window
Functions

Step 2: Querying to perform window functions:

QUERY

SELECT s_id,score, RANK() OVER (ORDER BY score DESC) my_rank , PERCENT_RANK()

OVER (ORDER BY score DESC) percentile_rank , ROW_NUMBER() OVER (ORDER BY
score) row_num, CUME_DIST() OVER (ORDER BY score) cume_dist_val FROM
marksheet;
Assisted Practice: Ranking and Miscellaneous Window
Functions

Output:
 Key Takeaways

Window functions provide an aggregated value for each row

with a unique identity.

Aggregate window functions perform on a particular set of

rows and provide the result in a single row.

Ranking window functions specify the rank for individual fields

as per the categorization.
Knowledge Check
Knowledge
Check
What is the result of Window Functions ?
1

A. Aggregate value of each row

B. Group of values

C. Sorted values

D. Divides into partition

 Knowledge
Check
What is the result of Window Functions ?
1

A. Aggregate value of each row

B. Group of values

C. Sorted values

D. Divides into partition

The correct answer is A

Window functions perform various operations on a group of rows and provide an aggregated value for each row.
Knowledge
Check
Which of the following are the clauses in MySQL ?
2

A. Row number and Rank

B. Last value and First value

C. Minimum and Maximum

D. Partition, Frame, and Order By

 Knowledge
Check
Which of the following are the clauses in MySQL ?
2

A. Row number and Rank

B. Last value and First value

C. Minimum and Maximum

D. Partition, Frame, and Order By

The correct answer is D

The types of clauses are Partition Clause, Frame Clause, and Order By Clause.
Knowledge
Check
What is the return type in COUNT() ?
3

A. DECIMAL

B. FLOAT

C. BIGINT

D. DOUBLE
 Knowledge
Check
What is the return type in COUNT() ?
3

A. DECIMAL

B. FLOAT

C. BIGINT

D. DOUBLE

The correct answer is C

The return type of COUNT() is BIGINT.

Knowledge
Check
Which ranking window function returns a value from zero to one ?
4

A. NTH value

B. Percent rank

C. N title

D. Row number
 Knowledge
Check
Which ranking window function returns a value from zero to one ?
4

A. NTH value

B. Percent rank

C. N title

D. Row number

The correct answer is B

Percent Rank returns value from zero to one.

Knowledge
Check
What is the purpose of LEAD Function ?
5

A. Retrieve the values from next N rows

B. Retrieve the values from previous N rows

C. Retrieve unique sequential number

D. Assign rank to all the rows within every partition

 Knowledge
Check
What is the purpose of LEAD Function ?
5

A. Retrieve the values from next N rows

B. Retrieve the values from previous N rows

C. Retrieve unique sequential number

D. Assign rank to all the rows within every partition

The correct answer is A

LEAD function retrieves the values from next N rows.