Transcript Summer Class 3

Business Analysis
ITEC-630 Fall 2010
Information & Data Analysis
Professor J. Alberto Espinosa
1
Agenda
• Introduction to database concepts
• Data modeling & relational database design
• Transitional artifacts: the CRUD matrix –
linking requirements to data design
• Normalization
2
The Big Picture
BPM/UC
Matrix
Business
Process
Analysis
CRUD
Matrix
System
Functional &
Non-Functional
Requirements
Analysis
Information
Analysis
Figure out the data requirements
to support the application
3
Data Modeling
Concepts
4
How Most Business Applications
are Implemented
Business
Application 1
Business
Application 2
Business
Application 3
Etc
Database Management System (i.e., Database Platform)
(e.g., Oracle, Access, SQL Server, etc.)
Database 1
Database 2
Database 3
Database 4
Etc.
5
Stand-alone DBMS
DBMS and database work in the same computer:
the user’s computer  OK for personal productivity
Stand-alone
DBMS
(e.g., MS Access)
Database
6
DBMS in a Client/Server Environment:
Better for corporate use  the DBMS has two components
DBMS Server: runs the “back-end” part of the DBMS and
performs most of the data management functions – e.g.,
queries, updates, etc.
DBMS Client: runs “front-end” part of the DBMS that provides the
user interface (e.g., data entry, screen displays or presentation,
report formatting, query building tools)
DBMS
Client
Data Request
(e.g., query)
Response
(e.g., query result)
DBMS
Server
Database
Retrieve, add,
delete and/or
update data
7
DBMS in a Web Server Environment:
Very common when there are large numbers of users and would be impractical to
deploy and install a DBSM client  access to the database is done
through a browser (e.g., on-line purchases)
Request (ex. get a price quote, place an order)
Response (ex. query results with HTML-formatted
product price or order confirmation notice)
8
Business to Business
E-Commerce Example
using XML
e.g., supplier
INSERT
query
XML
Processor
DBMS
(e.g., MS
SQL Server)
XML Document
(e.g., Purchase
Order)
Internet
XML
Processor
e.g., buyer
SELECT
query
XML Document
(e.g., Purchase
Order)
DBMS
(e.g., Oracle)
9
Most Common Database Models
•
•
•
•
Hierarchical (of historical interest only)
Network (of historical interest only)
Relational
Object Oriented (new)
10
Relational Database
• For a database to be truly relational, it must comply
with 12 rules defined by its inventor (Dr. E. F. Codd).
• No commercially available database complies with
the full set of rules, but the 12 rules are used as
guidelines for sound database design.
• Rule 1 states that data should be presented in tables
• Rule 2 states that data must be accessible without
ambiguity
• We will talk more about other rules later (i.e., about
entity integrity and referential integrity – stay tuned).
11
Implications about Rule 1
A relational database must have:
•
•
•
Tables: or “entities”
Every table has a unique name
Ex. Students, Courses
Fields: or “columns”, “attributes”
Every field has a unique name within the table
Ex. Students (StudentID, StudentName, Major, Address)
Ex. Courses (CourseNo, CouseName, CreditPoints,
Description)
Records: or “rows”, “tuples”, “instances”
Every record is unique (has a unique field that identifies it)
Ex. {“jdoe”, “John Doe”, “CS”, 5000 Forbes Ave.)
Ex. {“MGMT-352-001”, “MIS”, Fall 2002, “A great course”}
12
Object Oriented (OO) Databases
•
•
•
•
•
•
OO languages + added database functionality, or
Database products + added OO programming facilities
Similar to relational databases
“Classes” (a grouping of similar objects -- like tables)
“Objects” (an instance of a class -- like records)
“Object properties” (object attributes -- like fields)
• Plus:
– Methods (i.e., procedures or programs)
Programs embedded in classes and objects
– Other OO Properties (inheritance, encapsulation, etc.)
13
Terminology Equivalence
ERD or
Data Model
OO Database
Relational
Database
Entity
Class
Table
Instances
Objects
Records
Relationship
Relationship
Relationship
Attributes
Properties
Fields
Other
Terms Used
Rows, Tuples
Columns
14
Important Data
Modeling Concepts
15
Data Modeling Goals
•
•
•
•
•
Data integrity
Avoid anomalies in the data
No data redundancy
Record the data in one place only
Efficient data entry
Duplicate data means having to enter the same data more than
once
Consistency
Duplicate data can lead to inconsistencies when the data changes
e.g., 2 different addresses for same client
Flexibility and easy evolution
East to maintain, update and add new tables
16
Data Integrity Issue #1:
Enforcing Entity Integrity
 Inspect Each Table
17
Entity Integrity
•
•
Is ensuring that every record in each table in the database can
be addressed (i.e., found) – this means that there each
record has to have a unique identifier that is not duplicate
or null (i.e., not blank)
Examples: every student has an AU ID; every purchase order
has a unique number; every customer has an ID
Primary key (PK)  helps enforce Entity Integrity:
• Field(s) that uniquely identifies a record in a table
(e.g., AU user ID)
• Entity integrity = PK is not duplicate & not blank
• PK can be:
– A single field (e.g., UserID), or
– Or more than one field (e.g., OrderNo, LineItem)
18
Data Integrity Issue #2:
Enforce Referential Integrity
 Inspect each relationship between any two tables
19
Referential Integrity
• Is ensuring that the data that is entered in one table
is consistent with data in other tables
• Examples: purchase orders can only be placed by
valid customers; accounting transactions can only be
posted to valid company accounts
Foreign key (FK)  helps enforce referential Integrity:
• A field in a table that is a PK in another table
• That is, a field that “must” exist in another table
• This is how referential integrity is maintained
20
Illustration: Primary and Foreign Keys
PK
FK
PK
21
Entity, Referential Integrity
PK
PK
PK, FK
PK
FK
Database Schema:
The structure of the
database, which contain
tables, views, constraints,
relations, etc. – just about
everything, except the
data itself
PK, FK
22
Other Important Keys
•
Candidate Keys:
– Often there are more than one keys that could serve
as a primary key
– Example: Order, LineItem vs. Order, ProdID
– Example: AU ID, SSN, AU Login ID
– These are called candidate
– Any candidate can be selected as the primary key
•
Alternative Keys:
– Once a primary key has been selected from the
choice of candidate keys, the other keys (not used
as PKs) are referred to as “alternative keys”
23
Developing Data Models
also called Entity-Relationship Diagram (ERD)
24
Data Model Example
Course Registration System
Courses
Instructors
CourseNo
CourseDescription
Many
InstructorID
CreditPoints
PreRequisites
ClassroomNo
1
Includes
InstructorID
Teach
1
Relationships
StudentID
Enrollments
Comments
Entities
Students
Many
StudentID
CourseNo
LastName
FirstName
Telephone
EMailAddr
Enrolls
Many
1
LastName
FirstName
SSN
Department
College
Major
EMailAddr
25
Data Model Example (MS Access equivalent)
Course Registration System
Cardinality
1 to
Many
Enrolls
Entities
Relationships
26
The Textbook’s ERD Notation
Entities
InstructorID
LastName
CourseNo
FirstName
Teach
Instructors
Telephone
InstructorID
(FK)
EMail
CourseDescr
Courses
CreditPoints
PreReqs
Relationships
27
Peter Chen’s ERD Notation
Instructors
PK
Course
InstructorID
Teaches
LastName
FirstName
Telephone
EMail
PK
FK1
CourseNo
CourseDescription
InstructorID
CreditPoints
PreRequisites
28
Conceptual Data Modeling
• Data-oriented modeling method that describes the
data and relationships among data entities
• Goal: capture meaning of the data
• 2 main ERD or data model constructs:
 Entities and its attributes
 Relationships between entities
29
Entity
“An object, person, place, event or thing or which we want to
record data”
•
•
Equivalent to a table in a database
Examples: instructors, students, classrooms, invoices,
registration, machines, countries, states, etc.
•
Entity instance: a single occurrence of an entity
Example: Espinosa, KSB T58, ITEC 455
•
Entities can be identified in a requirements analysis
description by following the use of NOUNS
30
Relationships
• Relationships describe how two entities relate to each
other
• Relationships in a database application can be identified
following the VERBS that describe how entities are
associated with one another
• Examples:
students enroll in courses
countries have cities, etc.
31
Cardinality
•
Cardinality is an important database concept to describe how two
entities are related
•
The Cardinality of a relationship describes how many instances of
one entity can be associated with another entity
The cardinality of a relationship between two entities has two
components:
– Maximum Cardinality: is the maximum number of instances that
can be associated with the other entity – usually either 1 or many
(the exact number is rarely used)
– Minimum Cardinality: is the minimum number of instances that
can be associated with the other entity – usually either 0 or 1
– Symbols:
0
1
Many
•
32
Cardinality (cont’d.)
•
A relationship is fully described by describing the
cardinality in both directions of the relationship: e.g., a
client places zero (i.e., optional) or many orders and
each order must relate to only one (i.e., mandatory)
client.
•
Examples:
1 student can only park 1 (or 0) cars  1 to (0 or) 1
1 client can place (0 or ) many orders  1 to (0 or) many
1 student can enroll in (at least 1 or) many courses and
a course can have (0 or) many students  (0 or) many
to (1 or) many
33
Example: 2 Entities, 1 Relationship
Instructors
PK
Zero or
many
InstructorID
Teaches
LastName
FirstName
Telephone
EMail
One and
only one
Course
PK
FK1
CourseNo
CourseDescription
InstructorID
CreditPoints
PreRequisites
Peter Chen’s notation
& MS Visio software
34
ERD SYMBOLS (cont’d.)
Note: high level conceptual models don’t show attributes, just entities
Employee
BioData
Has
1 to 1
Maximum
Cardinality
(outer symbol)
Employee
Has
Mandatory
FamilyData
Optional
Minimum
Cardinality
(inner symbol)
Peter Chen’s notation
using Systems Architect software
35
ERD SYMBOLS (cont’d.)
→ Advises
← Have
Advisor
Student
1 to Many
Maximum
Cardinality
Faculty
1 to Many (or None)
Mandatory
Teaches
Minimum
Cardinality
Course
Optional
Peter Chen’s (“crow’s feet”) notation
using Systems Architect software
36
Many to Many Relationships?
Many to Many
Orders
Products
Convert a Many-to-Many
into 2 One-to-Many’s
Orders
Products
1 to Many
LineItems
1 to Many
Intersection Table (or None)
37
Cardinality: 1 to 1 (MS Access
notation)
38
Cardinality: 1 to many
(MS Access notation)
39
Steps in data modeling Modeling
1.
Identify and diagram all ENTITIES
2.
Add PK attributes – i.e., implement entity integrity
Ensure PK’s are non-null & non-duplicates
3.
Identify and diagram all RELATIONSHIPS
Note CARDINALITIES (1 to 1, 1 to n, n to n)
4.
Add FK attributes – i.e., implement referential
integrity (this is automatic in some tools—MS
Access)
5.
Add remaining attributes
40
ERD Example:
Course Registration System
Courses (CourseNo (PK), CourseDescripition, InstructorID,
CreditPoints, ClassroomNo)
PreRequisites (CourseNo (PK), PreRequisiteNo (PK),
Comments)
Students (StudentID (PK), LastName, FirstName, SSN,
Department, College, Major, EMail)
Enrollment (StudentID (PK), CourseNo (PK), Comments)
Instructors (InstructorID (PK), LastName, FirstName,
Telephone, EMail)
Classrooms (ClassroomNo (PK), ClassroomName, Building,
BuildingRoomNo, Equipment, Capacity)
Note: PK denotes a primary key
41
Example: Course Registration System
Step 1. Draw Entities
PreRequisites
Course
ClassRooms
Enrollment
Instructors
Students
42
Example: Course Registration System
Step 2. Add PK’s (undeline/separate with a line)
PreRequisites
Course
Instructors
CourseNo
PreRequisiteNo
CourseNo
InstructorID
Enrollment
ClassRooms
ClassroomNo
StudentID
CourseNo
Students
StudentID
43
Example: Course Registration System
Step 3. Add Relationships (w/Cardinalities)
PreRequisites
PK,FK1
PK
CourseNo
PreRequisiteNo
has
Course
PK
Instructors
Teaches
CourseNo
PK
InstructorID
Includes
Assigned
Enrollment
ClassRooms
PK
ClassroomNo
PK,FK1
PK,FK2
StudentID
CourseNo
Students
Enrolls
PK
StudentID
44
Example: Course Registration System
Step 4. Add FK’s
PreRequisites
PK,FK1
PK
CourseNo
PreRequisiteNo
Assigned
Course
has
PK
CourseNo
FK1
FK2
InstructorID
ClassroomNo
Instructors
Teaches
PK
InstructorID
Includes
Enrollment
ClassRooms
PK
ClassroomNo
PK,FK1
PK,FK2
StudentID
CourseNo
Students
Enrolls
PK
StudentID
45
Example: Course Registration System
Step 5. Add Remaining Attributes
Course
PreRequisites
PK,FK1
PK
CourseNo
PreRequisiteNo
Comments
Assigned
Has
PK
CourseNo
FK1
CourseDescription
InstructorID
CreditPoints
FK2
ClassroomNo
Instructors
Teaches
PK
InstructorID
LastName
FirstName
Telephone
EMail
Students
ClassRooms
PK
Includes
PK
ClassroomNo
ClassroomName
Building
BuildingRoomNo
Equipment
Capacity
Enrollment
PK,FK1
PK,FK2
StudentID
CourseNo
Comments
Enrolls
StudentID
LastName
FirstName
SSN
Department
College
Major
EMail
46
Example:
Course Registration System
(in MS Access)
47
EXAMPLE:
Package Delivery Tracking System
Deliveries
Clients
PK
ClientID
Deliveries
Deliveries
Deliveries
PK
PK DeliveryNo
DeliveryNo
PK
PK DeliveryNo
DeliveryNo
LastName
FirstName
Address
Telephone
FK4 ClientID
ClientID
FK4
FK5 DriverNo
FK5 DriverNo
Date
Status
Clients
Clients
PK
PK
ClientID
Trucks
Trucks
PK
PK
Trucks
PK
TruckNo
TruckNo
Packages
Packages
Packages
Packages
PackageNo
PK
PK
PK PackageNo
PackageNo
PackageNo
FK4
DeliveryNo
FK4 DeliveryNo
Size
Charge
Drivers
Drivers
Drivers
PK
PK DriverNo
DriverNo
PK
DriverNo
Drivers
FK1
TruckNo
TruckNo
PK
DriverNo
Make
Model
Year
FK1
TruckNo
DriverName
LicenseNo
48
Example:
Package Delivery Tracking System
49
EXAMPLE:
Airline Reservation System
50
Example:
Airline Reservation System
51
Final Data Modeling Step:
“Normalize” Your Design
(we will discuss this later)
52
Transitional Artifact:
The CRUD Matrix 
Connecting Data
Objects to Use Cases
53
Identifying Data Entities from Use Cases
• Identify and highlight (or bold face) all nouns in the use
cases
• Inspect these nouns to see if they represent possible data
entities (i.e., database tables)
• But be careful, a noun may not refer to an entity, but simply
to an attribute of an entity
 A data entity is something you want to collect data
about (e.g., Students)
 An attribute is the data you want to collect about that
entity (StudentID, Name, SSN, EmailAddress)
54
The CRUD Matrix
• A “transitional artifact” is one that helps establish a relationship or
cross reference between artifacts
• A CRUD matrix is a transitional artifact between Use Cases and
Data Entities
• Helps ensure that the Use Cases specified have all the necessary
Data Entities to handle the data needs of the application and,
conversely, that the set of Data Entities identified cover the entire
functionality specified in the requirements.
• The Use Cases, if properly specified, must describe all the actions
necessary to maintain all the application’s database tables
• A CRUD matrix is a table that cross references which Use Cases:
(C)reate, (R)ead, (U)pdate and/or (D)elete data in these objects
55
Developing a CRUD Matrix
• The CRUD matrix has one row for every data entity identified and
one column for every Use Case specified (or the other way around)
• So, first create a column (or row) for every Use Case in your model
• Every noun highlighted in the Use Cases will suggest the need for
data entity to store the respective data you, so you need to create a
row (or column) for each of these data entities
• Then go through every cell in the first Use Case and enter a C, R, U
and/or D on the cell depending on whether the Use Case is creating,
reading, updating or deleting records in the respective data entity (i.e.,
database table).
• The C’s, R’s, U’s and D’s should give you an idea of the SQL queries
that you will need to develop for your application
56
Illustration
Entity 1
Entity 2
Entity 3
UC-101
UC-102
C
R
UC-103
U
D
• UC-102 Reads data from Table 1
 It will require an SQL SELECT query
• UC-101 Creates a record in Table 1
 It will require an SQL INSERT query
• UC-103 Deletes records data from Table 3
 It will require an SQL DELETE query
• UC-102 Updates data in Table 2
 It will require an SQL UPDATE query
57
CRUD Matrix Example for a Loan
Processing Application
Use Case
Data Entity
Submit a Loan
Request
Evaluate a Loan
Request
Applicant
C
Loan Application
C
R
Credit Score
C
R
Credit Report
C
R
Account History
C
R
Loan Request
C
R,U
Loan Officer
R
Evaluation
C
Book a Loan
R
R
Loan Agreement
R
Loan Account
C
Loan Clerk
R
In a database application, these are tables and these are queries
58
ATM Application Example
59
ATM Use Case
Use Case ID
UC-100
Use Case
Withdraw Funds
Actors
(P) Customer
Description
The customer inserts card in the ATM, logs in with a pass code, and
makes a selection from the available choices to withdraw funds. Once in
the funds withdrawal screen, the customer is prompted to enter the
amount to withdraw. After the amount is entered, the system will check
for availability of funds for that customer. Provided that funds are
available, the system will dispense the amount requested in cash and
then debit that amount from the customer’s bank account. The system
will record the last withdrawal date in customer’s file and record
transaction in ATM transaction log .
Priority
Non-Functional
Requirements
Assumptions
Source
60
ATM Use Case
Use Case ID
UC-101
Use Case
Deposit Funds
Actors
(P) Customer
Description
The customer inserts card in the ATM, logs in with a pass code, and
makes a selection from the available choices to deposit funds. Once in
the funds deposit screen, the customer is prompted to enter the amount
to deposit. After the amount is entered, deposit slot door opens, customer
places deposit envelop in slot, deposit slot door closes. The system
credits the customer’s account accordingly, records the last deposit
date in the customer’s file and record the transaction in ATM
transaction log.
Priority
Non-Functional
Requirements
Assumptions
Source
61
ATM Use Case
Use Case ID
UC-102
Use Case
Transfer Funds
Actors
(P) Customer
Description
The customer inserts card in the ATM, logs in with a pass code, and makes
a selection from the available choices to transfer funds. Once in the funds
transfer screen, the customer is prompted to enter the amount to transfer,
from account and to account. After the information is entered, the checks
for availability of funds. If funds are available, it displays the transaction
and asks for confirmation. The customer confirms transaction and the
customer’s account gets adjusted accordingly. The system records the last
funds transfer date in the customer’s file and records the transaction in
ATM transaction log.
Priority
Non-Functional
Requirements
Assumptions
Source
62
ATM Use Case
Use Case ID
UC-103
Use Case
Balance Inquiry
Actors
(P) Customer
Description
The customer inserts card in the ATM, logs in with a pass code, and
makes a selection from the available choice to inquire balances. The
machine prints balances, records the last balance inquiry date in the
customer’s file and records the transaction in ATM transaction log .
Priority
Non-Functional
Requirements
Assumptions
Source
63
ATM System’s CRUD Matrix
Use Case
Deposit
Funds
Transfer
Funds
Inquire
Balances
C
U
U
U
Customer File
R,U
R,U
R,U
R,U
Customer Account
R,U
U
R,U
R
C
U
U
Data Entity
ATM
ATM Transaction Log
Customer Transactions
Withdraw
Funds
R,U
64
Database Design Issue #5:
“Normalize” Your Design
65
Database Design Goals
•
Data integrity (Entity and Referential Integrity – ERD’s)
Avoid anomalies in the data
•
No data redundancy
Record the data in one place only
•
Efficient data entry
Duplicate data means having to enter the same data more than once
•
Consistency
Duplicate data can lead to inconsistencies when the data changes
e.g., 2 different addresses for same client
•
Flexibility and easy evolution
East to maintain, update and add new tables
Normalization
66
Why Normalization?
• Question: if a data model/ERD is sound and all entity integrity,
referential integrity, update/delete and business rules have been well
implemented, does this guarantee a good database design?
Answer: not necessarily. If your design is not “normalized”, you
could have redundant data, and that would be a BAD thing (design)
• Normalization should yield the most efficient way to organize and
record the data internally—not necessarily how users want to see the
data, but what makes more sense for non-redundant data storage
• We can later build user table views (i.e., what the user wants or
needs to see) by querying these normalized tables.
• Redundancy: only PK and FK (e.g., client ID’s) values should appear
in multiple tables (because they are needed to link tables)
 Non-key data (e.g., client last name) that appears in multiple tables is
“redundant”
67
Example
You gather requirements from users and one user gives you this table
and tell you that she would like the system to collect this data.
How would you organize this data internally in the database?
68
Normalization
•
Normalization = The systematic process of “decomposing” a set
of unorganized tables with redundant data into smaller, simpler,
and more organized tables with only minimal data redundant in key
fields and no data redundancy on non-key fields
— i.e., from chaos to order
Decompose to most efficient
internal organization 
You can always recover the original
data format with a query 
Decomposition
Query
69
Degree of Normalization
•
Normalization is a matter of degree -- the more normalized your design
is, the lower the chances of having redundant data
•
Normal Forms (NF) (higher NF designs are more normalized):
1NF  2NF  3NF  BCNF  PJNF  DKNF  4NF  5NF
•
The process of normalizing a design to 3NF may seem complex, but
the concept is very simple:
(1) Minimize data redundancy in key attributes
-- i.e., data in key fields can be entered in more than one table
(2) Eliminate data redundancy in non-key attributes
-- i.e., data in non-key fields should be entered only in one table
(3) Ensure that every piece of data (each non-key attribute)
can be unambiguously located by its PK
(4) Each incremental NF gets us a step closer in this direction
70
Normal Forms
To what extent is a database normalized?
• Normalization is a matter of degree
• Measured in what is called “normal forms” (NF)
• 1NF, 2NF, 3NF, etc., higher NF = more normalized
• 3NF
Good enough for most applications
• BCNF  Boyce-Codd NF (more robust version of 3NF)
Mostly of academic interest (and complex applications):
• 4NF, 5NF or PJNF (Project Join), DKNF (Domain-Key)
 More advanced theoretically, little practical use
 Useful for research and formal methods only
71
Q: What’s wrong with this table?
A: Data in PayDate & Amount fields not single-valued
—i.e., they have repeating values
72
Similar Table, Same Problem
A: repeating values for a PK value  PK is duplicate
73
First Normal Form (1NF)
•
•
•
A “TABLE” is in 1NF if
there are no multi-valued attributes
and no PK is duplicated
i.e., attributes are “atomic”
A “DATABASE” is in 1NF if
ALL its tables are in 1NF
74
Decomposition to 1NF:
Create a separate table where the repeating
values can be recorded as rows
75

Decomposition
76
Q: What’s wrong with this table?
A: Some data in the Client and OrderDate fields are entered twice
i.e., some non-key data are redundant
i.e., there are “partial dependencies” in the table (see next slide)
77
Functional Dependencies
• An attribute B is functionally dependent
on attribute A if the value of a valid
instance of attribute A uniquely
determines the value of attribute B
• Represented as:
A
B
78
Functional Dependency Examples
StudentID
StudentID
StudentName
StudentMajor
What are the functional dependencies in this relations?
Clients (ClientID, ClientName, City, State, Zip)
LineItems (OrderNo, LineItem, ClientID, ProdID, Qty)
79
Second Normal Form (2NF)
• Applies to tables with “composite” PKs
(i.e., PK has more than one attribute)
• A “TABLE” is in 2NF if
(1) it is in 1NF, and
(2) non-key attributes are functionally dependent
on the whole PK, not on just part of it (i.e., no partial dependencies)
• Note: we only need to worry about 2NF when PK contains more than
one attribute (i.e., “composite”)
• That is: if a table is in 1NF and has a single PK, it is automatically in
2NF
• A “DATABASE” is in 2NF if ALL its tables are in 2NF
80
Decomposition to 2NF:
Move the partial key (e.g., OrderNo) and the fields
that are functionally dependent on only that part of the
key (e.g., ClientID, OrderDate) to a separate table
and make that partial key the PK in that new table
81
Decomposition

82
Q: What’s wrong with this table?
A: Some of the data in the ClientCity field is redundant,
because once we know who the ClientID is, we know the
city where they live
i.e., there are “transitive dependencies” in the table
83
Transitive Dependencies
•
If a non-key attribute C is functionally dependent on
another non-key attribute B (BC) and
B is in turn dependent on the PK attribute A (AB)
this implies C is transitively dependent on A (AC)
(through B or ABC), which will cause redundancies
•
In 2NF, all non-key attributes are functionally
dependent on the PK
•
Thus, in a 2NF table, a transitive dependency will occur
every time there is a functional dependency between
any two non-key attributes.
84
Transitive Dependency Examples
OrderNo
CourseNo
ClientID
InstructorID
ClientName
InstructorName
Are there transitive dependencies in these relations?
LineItems (OrderNo, LineItem, ProdID, Qty)
LineItems (OrderNo, LineItem, ProdID, ProdName, Qty)
85
Third Normal Form (3NF)
• A “TABLE” is in 3NF if (1) it is in 2NF and (2) nonkey attributes depend on the PK and nothing else
• That is, non-key attributes are NOT functionally
dependent on other non-key attributes
(just on the PK)
• In other words, there are no transitive
dependencies
• A “DATABASE” is in 3NF if ALL its tables are in
3NF
86
Decomposition to 3NF:
Move the fields with transitive
dependencies to a separate table
87
Decomposition

88
In Summary
• 1NF = no multi-value attributes (or no PK duplicates)
• 2NF = 1NF + the “whole” PK, not just part of it
• 3NF = 2NF + the PK and “nothing but” the PK
• Important! it is OK to have non-normalized designs,
and some database applications may actually
require a non-normalized design, but you must have
an understanding of which normalization form you
are violating and a good reason for doing it
89
Exercises
Indicate the normal form (PK underlined) and decompose to 3NF
Class (CourseNo, SectionNo, RoomNo)
Class (CourseNo, SectionNo, RoomNo, Capacity)
Class (CourseNo, SectionNo, CourseName, RoomNo, Capacity)
90
Exercises
POS System:
Indicate the normal form (PK underlined) and decompose to 3NF
Sales (SaleNo, ClientID, ClientName, SaleDate, SaleAmount)
SalesDetails (SaleNo, LineItem, SaleDate, ProdID, ProdName, Qty)
Other Systems:
VideoRental (VideoNo, Date, MovieID, MovieName, ClientID)
VideoRental (VideoNo, Date, ClientID, RentalDays)
Videos (VideoNo, MovieID, MovieName, MovieType)
Videos (VideoNo, MovieID, VideoCondition)
Movies (MovieID, MovieName, MovieType, Producer, ReleaseDate)
91
Exercise
Indicate the normal form and decompose to 3NF
92
Decomposition  Queries
Conceptually, normalization can be thought of the opposite of a
SELECT SQL query. When you normalize, you decompose a large
table into simpler, smaller tables without redundancies. In contrast,
when you query several small tables, the result is a larger table in
which redundancies don’t matter.
For example, the decomposed tables of the exercise in the prior page
can be reconstructed by querying the normalized tables as follows:
SELECT Companies.CompanyID, CompanyName,
Employees.EmployeeID, EmployeeName,
Departments.DeptID, DeptName
FROM Departments, Companies, Employees
WHERE Companies.CompanyID = Employees.CompanyID
AND Departments.DeptID = Employees.DeptID
93
Exercise
Indicate the normal form and decompose to 3NF
(and then try to write an SQL query to re-construct the original table)
94
Back to Basics:
Enterprise Architecture
Business
Domain
Business
Application
ITEC 630:
•Business Process
•Business Data Model
• Business Application Model
•Technology Infrastructure
Enterprise Process Model
Enterprise Data Model
Organization’s
Goals
Enterprise Application Model
Enterprise Technology Model
Enterprise Model
95