NESTED QUERIES

AND
AGGREGATION

CHAPTER 5 (6/E)
CHAPTER 8 (5/E)

1
LECTURE OUTLINE
 More Complex SQL Retrieval Queries
• Self-Joins
• Renaming Attributes and Results
• Grouping, Aggregation, and Group Filtering
• Ordering Results
• Nested SPJ Queries

2
REVIEW OF SPJ QUERIES IN
 SPJ (select-project-join) queries
• SQL’s basic select-from-where queries
• Equivalent to using only , , and ⋈ (or ) in Relational Algebra
(and possibly , if attributes need to be renamed before joining)

3
RENAMING IN
 For convenience, include renaming (like ) as well
 Aliases or tuple variables
• Provide alternative names for tables or columns

Customer Sale LineItem

custid name address phone saleid date custid saleid product quantity price

SELECT name, sale_date, product, quantity AS amount

FROM Customer C, Sale AS S(id,sale_date,custid), LineItem
WHERE C.custid = S.custid AND id = saleid;

• Keyword AS is optional

4
SELF-JOINS
 Renaming is mandatory if table used more than once in a query
EMPLOYEE
Fname Minit Lname Ssn Bdate Address Sex Salary Super_ssn Dno
 Example
Give the last names and salaries of employees and their managers whenever
the employee earns more than the manager.

• Think of the EMPLOYEE table as two tables, one for employees and one for
managers.
E
Fname Minit Lname Ssn Bdate Address Sex Salary Super_ssn Dno

M
Fname Minit Lname Ssn Bdate Address Sex Salary Super_ssn Dno

SELECT E.Lname, E.Salary, M.Lname, M.Salary

FROM EMPLOYEE E, EMPLOYEE M
WHERE E.Super_ssn = M.Ssn and E.Salary > M.Salary;

5
AGGREGATE FUNCTIONS
 Used to accumulate information from multiple tuples, forming a single- tuple
summary
 Built-in aggregate functions
• COUNT, SUM, MAX, MIN, and AVG
 Used in the SELECT clause
 Examples:
How many movies were directed by Steven Spielberg?
SELECT COUNT(*)
FROM Film
WHERE director='Steven Spielberg‘;
• All tuples in result are counted, with duplicates!
• COUNT(title) or COUNT(director) give same result!
• COUNT(DISTINCT year) would include each year only once!
What was the total movie profit since 2010, across how many directors?
SELECT SUM(gross - budget), COUNT(DISTINCT director)
FROM Film
WHERE year >= 2010;

6
GROUPING BEFORE
 How can we answer a query such as
“How many films were directed by each director after 2001?”
• Need to produce a result with one tuple per director
1. Partition relation into subsets of tuples based on grouping
column(s)
2. Apply function to each such group independently
3. Produce one tuple per group
 GROUP BY clause to specify grouping attributes
SELECT director, COUNT(*)
FROM Film
WHERE year > 2001
GROUP BY director;
• Every selector in SELECT clause must be a grouping column or an
aggregation function
• e.g., SELECT director, year, COUNT(*)
would not be allowed unless also grouping by year i.e.,
GROUP BY director, year

7
HAVING
 After partitioning into groups, whole partitions can be discarded.
• Provides a condition on the grouped tuples

 Having clause cannot reference individual tuples within group

• Can reference grouping column(s) and aggregates only

 Contrast WHERE clause to HAVING clause

Note: As for aggregation, no GROUP BY clause means relation treated as one

group

9
ORDERING OF QUERY
 Final output of a query can be sorted by one or more column values
 Use ORDER BY clause
• Keyword DESC for descending order of values
• Optionally use keyword ASC for ascending order (default)
 Example Course
SELECT dept, term,
ORDERING OF QUERY
dept cnum instructor term
COUNT(DISTINCT instructor) AS num_instructors
FROM Course
GROUP BY dept, term;
ORDER BY dept, term DESC;

10
ORDERING OF QUERY
• Note that this is sorted ascending by department.
• Within each department, terms sorted in descending order.
• What if DISTINCT omitted? What if term omitted from SELECT clause?
What if dept omitted from GROUP BY clause? What if dept omitted from
ORDER BY clause?
SUMMARY OF SQL QUERIES

1. Assemble all tables according to From clause (“,” means to use ).
2. Keep only tuples matching Where clause.
3. Group into blocks based on Group By clause.
4. Keep only blocks matching Having clause.
5. Create one tuple for each block using Select clause.
6. Order resulting tuples according to Order By clause.

1
 NESTED
 Any table can be used in FROM clause.
 select-from-where produces a table.
 Thus can nest one query within another.
 Example:
Give the biographical information for directors of profitable movies.
Film Person
title genre year director minutes budget gross name birth city
SELECT name, birth, city
FROM ( SELECT director
FROM Film
WHERE gross > budget) AS Profitable,
Person
WHERE director = name

12
NESTED QUERIES (CONT’D.)
 Any column can be used in SELECT and WHERE clauses.
• But refers to only one tuple value at a time
 select-from-where can produce a one-column table that
contains only one tuple.
 Thus queries can also be nested in SELECT and WHERE clauses
 Example:
Which film(s) had the highest budget?

SELECT *
FROM Film
WHERE budget = ( SELECT MAX(budget)
FROM Film);

14
USING IN FOR MEMBERSHIP
 Comparison operator IN
• Compares value v with a set (or bag) of values V
• Evaluates to TRUE if v is one of the elements in V
• Allows any relation in WHERE clause

• Can omit DISTINCT from this solution. Why?

15
USING IN
 Use tuples of values in comparisons
• Requires parentheses

16
NESTED 1-COLUMN
 Use other comparison operators to compare a single value v
• = ANY (or = SOME) operator
• Returns TRUE if the value v is equal to some value in the set V
• Equivalent to IN
• Also available for >, >=, <, <=, and <>
• >= ALL operator
• Returns TRUE if the value v is greater than or equal to every value in
the set V
• Equivalent to =(SELECT MAX(…)…)
• Also available for =, >, <, <=, and <>

17
CORRELATED NESTED
 Correlated nested query
• Evaluated once for each tuple in the outer query

 Such queries are easiest to understand (and write correctly) if all

column names are qualified by their relation names.
 Note that the inner query can refer to E, but the outer query cannot
refer to D.

18
 EXISTS AND UNIQUE FUNCTIONS
 [NOT] EXISTS function
• Check whether result of correlated nested query is empty or not
• EXISTS equivalent to (SELECT COUNT(*) … ) <> 0
Customer Sale
custid name address phone saleid date custid

SELECT name,
phone FROM
Customer C
WHERE NOT EXISTS ( SELECT *
FROM Sale S
WHERE C.custid = S.custid);

• Note that columns selected in inner query are irrelevant.

 SQL function UNIQUE(Q)

• Returns TRUE if no duplicate tuples in result of query Q

19
LECTURE SUMMARY
 Complex SQL:
• Self joins
• Aggregate functions
• Grouping
• Sorting
• Nested queries
 Relational algebra expressions can handle self joins and nested
queries with no additional operators
• Grouping, aggregations, and sorting require additional operators

21