0% found this document useful (0 votes)

22 views

SQL: Simple Analysis: Immanuel Trummer

The document discusses how to perform simple analysis on a database using SQL queries, explaining the basic SELECT, FROM, and WHERE clauses to retrieve data from one or more tables, join tables together, and filter rows, as well as more advanced features like aggregation queries, distinct results, and alternative join syntaxes. Sample queries are provided to illustrate these concepts and analyze student enrollment data from multiple tables.

Uploaded by

KrishanSingh

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

22 views

SQL: Simple Analysis: Immanuel Trummer

Uploaded by

KrishanSingh

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 37

SQL: Simple Analysis

Immanuel Trummer 
[email protected]

www.itrummer.org
Database Management
Systems (DBMS)
Application 1 Connections, Security, Utilities, ...

Application 2
DBMS Interface Query Processor
Query Parser Query Rewriter

... Query Optimizer Query Executor

Storage Manager
Data Access Buffer Manager

Transaction Manager Recovery Manager

Slides by Immanuel Trummer, Cornell University

Data
Reminder

• Learning SQL (Structured Query Language)

• Have seen how to define database schema

• Have seen how to insert/update/delete data

• Now: how can we analyze data via SQL queries?

Slides by Immanuel Trummer, Cornell University

Simple SQL Queries

• An SQL query describes a new relation to generate

• Simple SQL queries consist of three parts

• SELECT: describes columns of relation to generate

• FROM: describes source relations and how to match

• WHERE: defines conditions result rows must satisfy

Slides by Immanuel Trummer, Cornell University

Simple Query Format
• SELECT <columns> 
FROM <table1> JOIN <table2> ON (<join-pred>) ... 
WHERE <where-pred>

• <columns> is comma-separated list of columns

• <table1> and <table2> are database relations

• <join-pred> is condition defining matching tuples pairs

• <where-pred> are additional conditions

Slides by Immanuel Trummer, Cornell University

Example Query
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Students.Sname 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Example Query
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

Find pairs of students

and enrollment tuples where Sid (i.e.,
SELECT Students.Sname  student ID) is the same ...
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Example Query
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Students.Sname 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'
... pair that with courses
where Cid (i.e., course ID) matches
the one in enrollment ...
Slides by Immanuel Trummer, Cornell University
Example Query
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Students.Sname 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'
... filter to rows where Cname
(course name) is 'CS4320' ...
Slides by Immanuel Trummer, Cornell University
Example Query
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

... then discard all columns

except for Sname (student
SELECT Students.Sname  name).
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Exercise (5 Minutes)
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

• For the same database, find all names of courses for

which student "John Smith" is enrolled!

Slides by Immanuel Trummer, Cornell University

SELECT S.Sname 
FROM Students S 
JOIN Enrollment E ON (S.sid = E.sid)  
JOIN Courses C ON (E.cid = C.cid)  
WHERE C.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Assign alias S for

SELECT S.Sname  student relation ...
FROM Students S 
JOIN Enrollment E ON (S.sid = E.sid)  
JOIN Courses C ON (E.cid = C.cid)  
WHERE C.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Simplification by Alias
SELECT Students.Sname 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'
... use alias
instead of full table
SELECT S.Sname  name
FROM Students S 
JOIN Enrollment E ON (S.sid = E.sid)  
JOIN Courses C ON (E.cid = C.cid)  
WHERE C.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Omitting Table Names
SELECT S.Sname 
FROM Students S 
JOIN Enrollment E ON (S.sid = E.sid)  
JOIN Courses C ON (E.cid = C.cid)  
WHERE C.Cname = 'CS4320'
Can
omit table names if no
SELECT Sname  ambiguity exists.
FROM Students S 
JOIN Enrollment E ON (S.sid = E.sid)  
JOIN Courses C ON (E.cid = C.cid)  
WHERE Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

More Diverse Predicates
• Can use inequalities (>, >=): Students.scores > 70

• Writing "not equal": Courses.Cname <> 'CS4320'

• Check if value in list: Cname IN ('CS4320', 'CS5320')

• Regular expressions: Cname LIKE 'CS_320%'

• % stands for zero or more arbitrary characters

• _ stands for one arbitrary character

Slides by Immanuel Trummer, Cornell University

Composite Predicates

• Logical conjunction via AND keyword

• Logical disjunction via OR keyword

• Negation via NOT keyword

• E.g., Cname = 'CS4320' OR Cname = 'CS5320'

Slides by Immanuel Trummer, Cornell University

Diverse Select Clauses
• Shortcuts for selecting multiple columns

• * selects all columns

• <table>.* selects all columns from <table>

• Can use arithmetic expressions in select clause

• E.g., SELECT 3 * (<column1> + <column2>)

• Can assign new names for output columns

• E.g., SELECT Sname as StudentName

Slides by Immanuel Trummer, Cornell University

Join Syntax Alternatives
• Simply specify names of columns that appear in multiple tables

• <table1> JOIN <table2> USING (<column>)

• Abbreviates <table1> JOIN <table2>  

ON (<table1>.<column> = <table2>.<column>)

• "Natural joins" match values in columns with same name

• <table1> NATURAL JOIN <table2>

• Introduces equality conditions between columns of same name

• No join keyword: FROM <table1>, <table2> WHERE <join-condition>

Slides by Immanuel Trummer, Cornell University

Distinct Results

• SELECT ... may generate the same row multiple times

• Use SELECT DISTINCT instead to eliminate duplicates

Slides by Immanuel Trummer, Cornell University

Aggregation Queries
• Can calculate aggregates over all rows of result relation

• SQL Aggregates: COUNT, SUM, AVG, MIN, MAX

• SUM, AVG, MIN, MAX: numerical expression parameter

• COUNT(*) for counting rows in result relation

• COUNT(<column>) counts rows with value in <column>

• COUNT(DISTINCT <column>) counts number of

distinct values in <column> in result relation

Slides by Immanuel Trummer, Cornell University

Aggregation Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*) 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Aggregation Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

Find pairs of students

and enrollment tuples where Sid (i.e.,
SELECT Count(*)  student ID) is the same ...
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Aggregation Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*) 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'
... pair that with courses
where Cid (i.e., course ID) matches
the one in enrollment ...
Slides by Immanuel Trummer, Cornell University
Aggregation Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*) 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'
... filter to rows where Cname
(course name) is 'CS4320' ...
Slides by Immanuel Trummer, Cornell University
Aggregation Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

... then count all remaining rows.

SELECT Count(*) 
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Courses.Cname = 'CS4320'

Slides by Immanuel Trummer, Cornell University

Aggregation by Group

• Common: want aggregates for multiple data subsets

• Use SQL GROUP-BY clause to define data subsets

• GROUP BY <column-list> - distinguish data subsets

based on their values in specified columns

Slides by Immanuel Trummer, Cornell University

Grouping Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*), Cname 

FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname

Slides by Immanuel Trummer, Cornell University

Grouping Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

Find pairs of students

and enrollment tuples where Sid (i.e.,
SELECT Count(*), Cname 
student ID) is the same ...
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname

Slides by Immanuel Trummer, Cornell University

Grouping Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*), Cname 

FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname
... pair that with courses
where Cid (i.e., course ID) matches
the one in enrollment ...
Slides by Immanuel Trummer, Cornell University
Grouping Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*), Cname 

FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname
... filter to rows where
Cname (course name) is 'CS4320' or
'CS5320' ...
Slides by Immanuel Trummer, Cornell University
Grouping Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*), Cname 

FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname
... group remaining rows by
Cname (Course name) ...
Slides by Immanuel Trummer, Cornell University
Grouping Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)
... count rows in each
group and report count with course
SELECT Count(*), Cname  name (unique per group).
FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname

Slides by Immanuel Trummer, Cornell University

Grouping Details
• Grouping is applied after pairing data sources (FROM)
and filtering rows (WHERE)

• Result contains one row per group

• Implies restrictions on SELECT clause!

• Only expressions with unique value per group

• This includes aggregates and grouping columns

Slides by Immanuel Trummer, Cornell University

Predicates on Groups

• Condition in WHERE clause applies to single rows

(evaluated before grouping)

• HAVING clause specifies conditions on groups 

(evaluated after grouping)

Slides by Immanuel Trummer, Cornell University

Having Example
Database Relations: 
 
Students(Sid, Sname) 
Enrollment(Sid, Cid) 
Courses(Cid, Cname)

SELECT Count(*), Cname 

FROM Students  
JOIN Enrollment ON (Students.sid = Enrollment.sid)  
JOIN Courses ON (Enrollment.cid = Courses.cid)  
WHERE Cname IN ('CS4320', 'CS5320') 
GROUP BY Cname 
HAVING Count(*) >= 100 Only keep groups with
at least 100 rows!
Slides by Immanuel Trummer, Cornell University
Exercise