(SQL)

Aggregating Data
Using Group Functions

Asif Sohail
University of the Punjab
Punjab University College of Information Technology (PUCIT)

© 2009 Punjab University College of Information Technology (PUCIT) Slide 1

 What Are Group Functions?
• Group functions operate on sets of rows to give one
result per group.
EMP
DEPTNO SAL
--------- ---------
10 2450
10 5000
10 1300
20 800
20 1100
20 3000 “maximum MAX(SAL)
20 3000 salary in ---------
20 2975 the EMP table” 5000
30 1600
30 2850
30 1250
30 950
30 1500
30 1250

© 2009 Punjab University College of Information Technology (PUCIT) Slide 2

 Types of Group Functions
Function Description
MAX Returns the largest value
MIN Returns the smallest value
SUM Returns the sum of values
AVG Returns the average of values (ignoring NULL values)
COUNT Returns the count of NOT NULL values
MEDIAN Returns the middle value
STDEVE Returns the standard deviation (ignoring NULL values)
VARIANCE Returns the variation (ignoring NULL values)

© 2009 Punjab University College of Information Technology (PUCIT) Slide 3

 Using Group Functions

SELECT [column,] group_function(column)

FROM table
[WHERE condition]
[GROUP BY column]
[ORDER BY column];

© 2009 Punjab University College of Information Technology (PUCIT) Slide 4

 Using MIN and MAX Functions

• You can use MIN and MAX for any datatype.

SQL> SELECT MIN(hiredate), MAX(hiredate)

2 FROM emp;

MIN(HIRED MAX(HIRED
--------- ---------
17-DEC-80 12-JAN-83

© 2009 Punjab University College of Information Technology (PUCIT) Slide 5

 Using AVG and SUM Functions
• You can use AVG and SUM for numeric data.

SQL> SELECT AVG(sal), MAX(sal),

2 MIN(sal), SUM(sal)
3 FROM emp
4 WHERE job LIKE 'SALES%';

AVG(SAL) MAX(SAL) MIN(SAL) SUM(SAL)

-------- --------- --------- ---------
1400 1600 1250 5600

© 2009 Punjab University College of Information Technology (PUCIT) Slide 6

 Using the COUNT Function
• COUNT(*) returns the number of rows in a table
satisfying a given condition (if any).
SQL> SELECT COUNT(*)
2 FROM emp
3 WHERE deptno = 30;

COUNT(*)
---------
6

© 2009 Punjab University College of Information Technology (PUCIT) Slide 7

 Using the COUNT Function

• COUNT(expr) returns the number of rows

with non-null values for the expr.
SQL> SELECT COUNT(comm)
2 FROM emp
3 WHERE deptno = 30;

COUNT(COMM)
-----------
4

© 2009 Punjab University College of Information Technology (PUCIT) Slide 8

 Using the COUNT Function
• COUNT(DISTINCT expr) returns the number
of distinct non-null values of the expr.

SQL> SELECT COUNT(DISTINCT deptno)

2 FROM emp;

COUNT(DISTINCT deptno)
----------------------
4

© 2009 Punjab University College of Information Technology (PUCIT) Slide 9

 Group Functions and Null Values
• Group functions ignore null values in the
column.

SQL> SELECT AVG(comm)

2 FROM emp;

AVG(COMM)
---------
550

© 2009 Punjab University College of Information Technology (PUCIT) Slide 10

 Using the NVL Function
with Group Functions

• The NVL function forces group functions to

include null values.

SQL> SELECT AVG(NVL(comm,0))

2 FROM emp;

AVG(NVL(COMM,0))
----------------
157.14286

© 2009 Punjab University College of Information Technology (PUCIT) Slide 11

 Creating Groups of Data

EMP
DEPTNO SAL
--------- ---------
10 2450
10 5000 2916.6667
10 1300
20 800 “average DEPTNO AVG(SAL)
20 1100 salary
------- ---------
20 3000 2175 in EMP
20 3000 table 10 2916.6667
20 2975 for each 20 2175
30 1600 department” 30 1566.6667
30 2850
30 1250 1566.6667
30 950
30 1500
30 1250

© 2009 Punjab University College of Information Technology (PUCIT) Slide 12

 Creating Groups of Data:
GROUP BY Clause

SELECT column, group_function(column)

FROM table
[WHERE condition]
[GROUP BY group_by_expression]
[ORDER BY column];

• Divide rows in a table into smaller groups by

using the GROUP BY clause.

© 2009 Punjab University College of Information Technology (PUCIT) Slide 13

 Using the GROUP BY Clause
• All columns in the SELECT list that are not in group
functions must be in the GROUP BY clause.

SQL> SELECT deptno, AVG(sal)

2 FROM emp
3 GROUP BY deptno;

DEPTNO AVG(SAL)
--------- ---------
10 2916.6667
20 2175
30 1566.6667

© 2009 Punjab University College of Information Technology (PUCIT) Slide 14

 Using the GROUP BY Clause
• The GROUP BY column does not have to
be in the SELECT list.

SQL> SELECT AVG(sal)

2 FROM emp
3 GROUP BY deptno;

AVG(SAL)
---------
2916.6667
2175
1566.6667

© 2009 Punjab University College of Information Technology (PUCIT) Slide 15

 Grouping by More
Than One Column
EMP
DEPTNO JOB SAL
--------- --------- ---------
10 MANAGER 2450
DEPTNO JOB SUM(SAL)
10 PRESIDENT 5000
-------- --------- ---------
10 CLERK 1300
10 CLERK 1300
20 CLERK 800 “sum salaries in 10 MANAGER 2450
20 CLERK 1100 the EMP table
10 PRESIDENT 5000
20 ANALYST 3000 for each job,
20 ANALYST 6000
20 ANALYST 3000 grouped by
20 CLERK 1900
20 MANAGER 2975 department”
20 MANAGER 2975
30 SALESMAN 1600
30 CLERK 950
30 MANAGER 2850
30 MANAGER 2850
30 SALESMAN 1250
30 SALESMAN 5600
30 CLERK 950
30 SALESMAN 1500
30 SALESMAN 1250

© 2009 Punjab University College of Information Technology (PUCIT) Slide 16

 Using the GROUP BY Clause
on Multiple Columns

SQL> SELECT deptno, job, sum(sal)

2 FROM emp
3 GROUP BY deptno, job;

DEPTNO JOB SUM(SAL)

--------- --------- ---------
10 CLERK 1300
10 MANAGER 2450
10 PRESIDENT 5000
20 ANALYST 6000
20 CLERK 1900
...
9 rows selected.

© 2009 Punjab University College of Information Technology (PUCIT) Slide 17

 Illegal Queries
Using Group Functions

• Any column or expression in the SELECT list

that is not an aggregate function must be in
the GROUP BY clause.

SQL>
SQL> SELECT
SELECT deptno,
deptno, COUNT(ename)
COUNT(ename)
BYY c
cllaau
u ssee
22 FROM
FROM emp;
emp;
n t
t h
h GR O
ee G ROOUUPPB
m i
iissssinn g
g i
i n
o l u
u m
mnn m
Col
C
SELECT
SELECT deptno,
deptno, COUNT(ename)
COUNT(ename)
**
ERROR
ERROR at
at line
line 1:
1:
ORA-00937:
ORA-00937: not
not aa single-group
single-group group
group function
function

© 2009 Punjab University College of Information Technology (PUCIT) Slide 18

 Illegal Queries
Using Group Functions

– You cannot use the WHERE clause to restrict groups.

– You use the HAVING clause to restrict groups.

SQL>
SQL> SELECT
SELECT deptno,
deptno, AVG(sal)
AVG(sal)
s
see
22 FROM
FROM emp
emp l u
aau
33 WHERE AVG(sal)
AVG(sal) >> 2000 c
c l
WHERE 2000
R EE
R ss
44 GROUP
GROUP BY
BY deptno;
deptno; HEE
H uupp
e W
W r oo
h e
tth ctt g g r
WHERE AVG(sal) > 2000 s
see r
r iic
WHERE AVG(sal) > 2000 u u
tt reess t
t
** n
noo r
ann t
t o
o
ERROR
ERROR at
at line
line 3:Ca
3:C
ORA-00934: group function is not allowed here
ORA-00934: group function is not allowed here

© 2009 Punjab University College of Information Technology (PUCIT) Slide 19

 Excluding Group Results
EMP
DEPTNO SAL
--------- ---------
10 2450
10 5000 5000
10 1300
20 800
20 1100 “maximum DEPTNO MAX(SAL)
20 3000 salary --------- ---------
3000 per department
20 3000 10 5000
20 2975 greater than 20 3000
30 1600 $2900”
30 2850
30 1250
30 950
2850
30 1500
30 1250

© 2009 Punjab University College of Information Technology (PUCIT) Slide 20

 Excluding Group Results: HAVING Clause

• Use the HAVING clause to restrict groups

• Rows are grouped.
• The group function is applied.
• Groups matching the HAVING clause are displayed.

SELECT column, group_function

FROM table
[WHERE condition]
[GROUP BY group_by_expression]
[HAVING group_condition]
[ORDER BY column];

© 2009 Punjab University College of Information Technology (PUCIT) Slide 21

 Using the HAVING Clause

SQL> SELECT deptno, max(sal)

2 FROM emp
3 GROUP BY deptno
4 HAVING max(sal)>2900;

DEPTNO MAX(SAL)
--------- ---------
10 5000
20 3000

© 2009 Punjab University College of Information Technology (PUCIT) Slide 22

 Using the HAVING Clause

SQL> SELECT job, SUM(sal) PAYROLL

2 FROM emp
3 WHERE job NOT LIKE 'SALES%'
4 GROUP BY job
5 HAVING SUM(sal)>5000
6 ORDER BY SUM(sal);

JOB PAYROLL
--------- ---------
ANALYST 6000
MANAGER 8275

© 2009 Punjab University College of Information Technology (PUCIT) Slide 23

 Nesting Group Functions

• Display the maximum average salary.

SQL> SELECT max(avg(sal))

2 FROM emp
3 GROUP BY deptno;

MAX(AVG(SAL))
-------------
2916.6667

© 2009 Punjab University College of Information Technology (PUCIT) Slide 24

 Extended Group by Clauses
• ROLLUP extends the GROUP BY clause to
return a row containing a subtotal for each group
of rows, plus a row containing a grand total for
all the groups.
SELECT deptno, SUM(sal) SELECT deptno, SUM(sal)
FROM emp FROM emp
GROUP BY deptno; GROUP BY ROLLUP(deptno);

DEPTNO SUM(SAL) DEPTNO SUM(SAL)

30 9400 10 8750
20 10875 20 10875
10 8750 30 9400
- 29025

© 2009 Punjab University College of Information Technology (PUCIT) Slide 25

 Extended Group by Clauses
• Multiple columns can be passed to ROLLUP.
• The order of ROWS is significant.
SELECT deptno,job,SUM(sal) SELECT deptno,job,SUM(sal)
FROM emp FROM emp
GROUP BY ROLLUP(deptno,job) GROUP BY ROLLUP(job,deptno);
DEPTNO JOB SUM(SAL) DEPTNO JOB SUM(SAL)
10 CLERK 1300 10 CLERK 1300
10 MANAGER 2450 20 CLERK 1900
10 PRESIDENT 5000
30 CLERK 950
10 - 8750
- CLERK 4150
20 CLERK 1900
20 ANALYST 6000 20 ANALYST 6000

20 MANAGER 2975 - ANALYST 6000

20 - 10875 10 MANAGER 2450
30 CLERK 950 20 MANAGER 2975
30 MANAGER 2850 30 MANAGER 2850
30 SALESMAN 5600
- MANAGER 8275
30 - 9400
- - -
- - 29025
© 2009 Punjab University College of Information Technology (PUCIT) - - 29025 Slide 26
 Extended Group by Clauses
• CUBE extends the GROUP BY clause to return
rows containing a subtotal for all combinations of
the values in the columns, plus a row containing
the grand total.
• Unlike ROLLUP, the order of columns in a CUBE
is not significant.

© 2009 Punjab University College of Information Technology (PUCIT) Slide 27

 SELECT deptno, job,SUM(sal) SELECT deptno,job,SUM(sal)
FROM emp FROM emp
GROUP BY CUBE(deptno,job); GROUP BY ROLLUP(deptno,job)

DEPTNO JOB SUM(SAL) DEPTNO JOB SUM(SAL)

- - 29025 10 CLERK 1300
10 - 8750 10 MANAGER 2450
20 - 10875 10 PRESIDENT 5000
30 - 9400 10 - 8750
- CLERK 4150 20 CLERK 1900
10 CLERK 1300
20 ANALYST 6000
20 CLERK 1900
20 MANAGER 2975
30 CLERK 950
20 - 10875
- ANALYST 6000
30 CLERK 950
20 ANALYST 6000
30 MANAGER 2850
- MANAGER 8275
30 SALESMAN 5600
10 MANAGER 2450
30 - 9400
20 MANAGER 2975
30 MANAGER 2850 - - 29025
- SALESMAN 5600
30 SALESMAN 5600
- PRESIDENT 5000
10 PRESIDENT 5000
© 2009 Punjab University College of Information Technology (PUCIT) Slide 28
 Summary
SELECT column, group_function(column)
FROM table
[WHERE condition]
[GROUP BY group_by_expression]
[HAVING group_condition]
[ORDER BY column];

• Order of evaluation of the clauses:

– WHERE clause
– GROUP BY clause
– HAVING clause

© 2009 Punjab University College of Information Technology (PUCIT) Slide 29

 Thank you for your attention.

Asif Sohail

Assistant Professor
University of the Punjab
Punjab University College of Information Technology (PUCIT)
Allama Iqbal (Old) Campus, Anarkali
Lahore, Pakistan
Tel: +92-(0)42-111-923-923 Ext. 154
E-mail: asif@pucit.edu.pk

© 2009 Punjab University College of Information Technology (PUCIT) Slide 30