Normalization of Database Tables

Lecture 10

1
Learning Outcomes
 In this chapter, you will learn:
 What normalization is and what role it plays in the database design
process
 About the normal forms 1NF, 2NF, 3NF, BCNF and 4NF
 How normal forms can be transformed from lower normal forms to
higher normal forms
 That normalization and ER modeling are used concurrently to
produce a good database design
 That some situations require denormalization to generate
information efficiently

2
Database Tables and Normalization
 Normalization
 Process for evaluating and correcting table structures to
minimize data redundancies
 Reduces data anomalies

 Series of stages called normal forms:

 First normal form (1NF)
 Second normal form (2NF)
 Third normal form (3NF)

3
Database Tables and Normalization
 Normalization (continued)
 2NF is better than 1NF; 3NF is better than 2NF
 For most business database design purposes, 3NF is as high
as needed in normalization

4
How Normalization Supports DB Design
If ER Modeling is
performed well, use
normalization as
validation technique.

This lecture follows

Approach 1

5
The Need for Normalization
 Example:
 A company that manages building projects
 Charges its clients by billing hours spent on each contract
 Hourly billing rate is dependent on employee’s position
 Periodically, report is generated that contains information such as
displayed in Table 6.1

A few employees works for one Project Num : 15

project.
Project Name :
Evergreen
Employee Num : 101,
102, 103, 105

6
Tabular Representation of the Report
Format

7
A Sample Report Layout for the Company

8
The Need for Normalization (cont’d.)
 Structure of data set in Figure 6.1 does not handle data very
well
 Proj_Num intended to be primary key but it contains nulls

Null

9
The Need for Normalization (cont’d.)
 Structure of data set in Figure 6.1 does not handle data
very well
 Invite data inconsistencies. E.g., Job_Class value
 “Elect. Engineer” might be entered as “Elect. Eng” or “El. Eng.” or “EE”

 Display data redundancies

 Delete anomalies – Suppose that only one employees is associated
with a given project. If that employee leaves the company and the
employee data are deleted, the project information will also be deleted
 Insert anomalies – A new employee must be assigned to a project.
 Update anomalies – Modifying the Job_Class for employee no. 105
requires many alterations.
The Normalization Process
 Each table represents a single subject
 No data item will be unnecessarily stored in more than
one table
 All nonprime attributes in a table are dependent on the
primary key
 Normalization is three-step procedure

11
The Normalization Process (cont’d.)

12
The Normalization Process (cont’d.)
 Objective of normalization is to ensure that all tables are
in at least 3NF
 Higher forms are not likely to be encountered in business
environment
 Normalization works one relation at a time
 Progressively breaks table into new set of relations based
on identified dependencies
 Normalizing table structure will reduce data redundancies

13
The Concepts of Functional Dependencies

14
 Conversion to First Normal Form
 Relational table must not contain repeating groups

Multiple entries
for the same
project no.
(proj_num = 15)

15
Conversion to First Normal Form (cont’d.)
 Step 1: Eliminate the Repeating Groups
 Eliminate nulls: each repeating group attribute contains an
appropriate data value

 Step 2: Identify the Primary Key

 Must uniquely identify attribute value
 New key must be composed

 Step 3: Identify All Dependencies

 Dependencies are depicted with a diagram

16
First Normal Form (NF)

17
Conversion to First Normal Form (cont’d.)
 Dependency diagram:
 Depicts all dependencies found within given table structure
 Helpful in getting bird’s-eye view of all relationships among
table’s attributes
 Makes it less likely that you will overlook an important
dependency

18
Dependency Diagram for 1NF

Transitive dependency:
dependency exists between two
19 nonprime attributes
Conversion to First Normal Form (cont’d.)
 First normal form describes tabular format:
 All key attributes are defined
 No repeating groups in the table
 All attributes are dependent on primary key
 All relational tables satisfy 1NF requirements
 Some tables contain partial dependencies
 Dependencies are based on part of the primary key

20
Conversion to Second Normal Form
 Step 1: Make New Tables to Eliminate Partial Dependencies
 Write each key component on separate line, then write original
(composite) key on last line
 Each component will become key in new table

 Step 2: Assign Corresponding Dependent Attributes

 Determine attributes that are dependent on other attributes
 At this point, most anomalies have been eliminated

21
Second Normal Form

22
Conversion to Second Normal Form (cont’d.)
 Table is in second normal form (2NF) when:
 It is in 1NF and
 It includes no partial dependencies:
 No attribute is dependent on only portion of primary key

 Employee table contains transitive dependencies

 Dependency of one nonprime attribute on another
nonprime attribute
 Job_Class  Chg_Hour

23
 Conversion to Third Normal Form
 Step 1: Make New Tables to Eliminate Transitive Dependencies
 For every transitive dependency, write its determinant as PK for new
table
 Determinant: any attribute whose value determines other values
within a row

 Step 2: Reassign Corresponding Dependent Attributes

 Identify attributes dependent on each determinant identified in Step 1
 Name table to reflect its contents and function

24
Third Normal Form

25
Conversion to Third Normal Form
 A table is in third normal form (3NF) when both of the
following are true:
 It is in 2NF
 It contains no transitive dependencies

26
Improving the Design
 Table structures should be cleaned up to eliminate initial
partial and transitive dependencies
 Normalization helps eliminate data redundancies

27
Improving the Design
 Issues to address, in order, to produce a good normalized
set of tables:
 Evaluate PK Assignments

 Evaluate Naming Conventions

 Refine Attribute Atomicity

 Identify New Attributes

Atomic attribute
– an attribute that cannot be subdivided
(e.g., emp_name is not atomic)
- should break emp_name to LastName,
FirstName, MidName to improve querying
flexibility
28
 Improving the Design
 Issues to address, in order, to produce a good normalized
set of tables (cont’d.):
 Identify New Relationships (e.g., Employee and Project)

 Refine Primary Keys as Required for Data Granularity

 Maintain Historical Accuracy Data Granularity

– level of details
 Evaluate Using Derived Attributes e.g., assign_hours
(Daily total? Weekly total?
Monthly total?)

Are all job_charge_hr the

- Attribute calculated from other attribute (recap!) same for all the projects ?
Assign_charge
29
= Assign_hrs * Assign_charge_hr
The Completed Database

30
Higher-Level Normal Forms
 Tables in 3NF perform suitably in business transactional
databases
 Higher-order normal forms are useful on occasion
 Two special cases of 3NF:
 Boyce-Codd normal form (BCNF)
 Fourth normal form (4NF)

31
The Boyce-Codd Normal Form (BCNF)
 Most designers consider the BCNF as a special case of
3NF
 When table contains only one candidate key, the 3NF and
the BCNF are equivalent
 Table can be in 3NF and fail to meet BCNF
 In 3NF, there is no partial dependencies, nor does it contain
transitive dependencies
 In BCNF, every determinant in the table be a candidate key
 What happens when a nonkey attribute is the determinant
of a key attribute?

32
A Table in 3NF but not BCNF

A + B  C, D
CB

B is a prime (key)
attribute

33
Sample Data for BCNF Conversion

34
 Decomposition of BCNF

A + B  C, D
A + C  B, D
Use A + C as the
CB primary keys, so that
B becomes nonprime
attribute

Class_code Staff_ID
Stu_ID
35
Class_code Enroll_grade
Fourth Normal Form (4NF)
 Table is in fourth normal form (4NF) when both of the
following are true:
 It is in 3NF
 No multiple sets of multivalued dependencies (one key
determines multiple values of two other attributes and
these attributes are independent of each other)
 E.g., one employee can have many service entries and many
assignment entries

36
Tables with Multivalued Dependencies

37
Tables in 4NF

38
Normalization and Database Design
 Normalization should be part of the design process
 Make sure that proposed entities meet required normal
form before table structures are created
 Many real-world databases have been improperly
designed or burdened with anomalies
 You may be asked to redesign and modify existing
databases

39
Normalization and Database Design
 Difficult to separate normalization process from ER modeling
process
 ER diagram
 Identify relevant entities, their attributes and relationships
 Identify additional entities and attributes

 Normalization procedures
 Focus on characteristics of specific entities
 Micro view of entities within ER diagram

40
Initial ERD

 EMPLOYEE(EMP_NUM, EMP_LNAME, EMP_FNAME,

EMP_INITIAL, JOB_DESCRIPTION, JOB_CHG_HOUR)
 PROJECT( PROJ_NUM, PROJ_NAME)
 Transitive dependency
 JOB_DESCRIPTION  JOB_CHG_HOUR

41
Modified ERD After Removing Transitive
Dependency

 EMPLOYEE(EMP_NUM, EMP_LNAME, EMP_FNAME, EMP_INITIAL,

JOB_CODE)
 PROJECT( PROJ_NUM, PROJ_NAME)
 JOB(JOB_CODE, JOB_DESCRIPTION, JOB_CHG_HOUR)
42
Incorrect M:N Relationship

43
Final Contracting COMPANY ERD

 EMPLOYEE(EMP_NUM, EMP_LNAME, EMP_FNAME, EMP_INITIAL,

JOB_CODE)
 PROJECT( PROJ_NUM, PROJ_NAME, EMP_NUM)
 JOB(JOB_CODE, JOB_DESCRIPTION, JOB_CHG_HOUR)
 ASSIGNMENT(ASSIGN_NUM, ASSIGN_DATE, PROJ_NUM, EMP_NUM,
ASSIGN_HOURS, ASSIGN_CHG_HOUR, ASSIGN_CHARGE)
The Implemented Database

45
Denormalization
 Creation of normalized relations is important database
design goal
 Processing requirements should also be a goal
 If tables are decomposed to conform to normalization
requirements:
 Number of database tables expands

46
Denormalization (cont’d.)
 Joining the larger number of tables reduces system speed
 Conflicts are often resolved through compromises that
may include denormalization
 Defects of unnormalized tables:
 Data updates are less efficient because tables are larger
 Indexing is more cumbersome
 No simple strategies for creating virtual tables known as
views

47
Data-Modeling Checklist
 Data modeling translates specific real-world environment
into data model
 Represents real-world data, users, processes, interactions
 Data-modeling checklist helps ensure that data-modeling
tasks are successfully performed
 Based on concepts and tools learned in Part II (Lec 2 - 4)

48
49
50