ADBMS

This is just a brief overview of all the modules

Note: ERD, Normalization problems are not included in this document
➢ MODULE 1
Q1) What is a Database Management System (DBMS), and how does it
differ from a file-based system?
Answer:
• Definition of DBMS:
o A software system designed to manage databases and provide an
interface for users and applications to interact with data.
o Ensures efficient storage, retrieval, and manipulation of data.
• Comparison with File-Based Systems:
o Advantages of DBMS over File Systems:
▪ Reduced Data Redundancy: Maintains a single, consistent
data source.
▪ Improved Data Integrity: Enforces rules and constraints,
ensuring valid and consistent data.
▪ Enhanced Security: Access controls prevent unauthorized
access.
▪ Support for Concurrent Access: Allows multiple users to
access and modify data simultaneously.
▪ Backup and Recovery Mechanisms: Protects against data
loss.
o Limitations of File-Based Systems:
▪ Data Redundancy and Inconsistency: Storing duplicate
data increases the risk of discrepancies.
▪ Limited Query Capabilities: File systems lack
sophisticated querying features.
▪ Poor Scalability: Performance degrades with data growth.
▪ Security Concerns: Minimal access control options.

Page 1 of 71
Q2) What are the ACID properties of a transaction in a DBMS, and why
are they important?
Answer:
• ACID Properties:
o Atomicity:
▪ Ensures that a transaction is completed entirely or not at all.
▪ If any part of a transaction fails, the entire transaction is
rolled back.
o Consistency:
▪ Guarantees that a transaction transforms the database from
one valid state to another.
▪ Preserves database integrity.
o Isolation:
▪ Transactions execute independently of each other.
▪ Prevents concurrent transactions from interfering with each
other.
o Durability:
▪ Ensures that the effects of a committed transaction persist,
even in the case of system failures.
• Importance:
o Maintains data integrity and reliability.
o Ensures predictable database behavior during concurrent access
and failures.
o Critical for applications such as banking, where consistency and
reliability are paramount.
Q3) What is data normalization, and what are its key benefits?
Answer:
• Definition:
o A process of organizing data to minimize redundancy and
dependency.

Page 2 of 71
 o Structures data into tables following specific normal forms.
• Key Benefits:
o Reduces Data Redundancy: Eliminates duplicate data.
o Improves Data Integrity: Enforces data consistency.
o Facilitates Maintenance: Easier to update and manage.
o Optimizes Storage: Reduces unnecessary data storage.
o Enhances Query Performance: Speeds up data retrieval by
avoiding duplicate scans.
• Types of Normalization:
o First Normal Form (1NF): Ensures atomicity of data (no
repeating groups).
o Second Normal Form (2NF): Eliminates partial dependency.
o Third Normal Form (3NF): Removes transitive dependencies.
Q4) What are the main types of SQL commands, and how are they
categorized?
Answer:
• Categories of SQL Commands:
o Data Definition Language (DDL):
▪ Commands: CREATE, ALTER, DROP, TRUNCATE,
RENAME.
▪ Purpose: Define and manage database schema and structure.
o Data Manipulation Language (DML):
▪ Commands: SELECT, INSERT, UPDATE, DELETE.
▪ Purpose: Retrieve and modify data.
o Data Control Language (DCL):
▪ Commands: GRANT, REVOKE.
▪ Purpose: Manage access and permissions.
o Transaction Control Language (TCL):
▪ Commands: COMMIT, ROLLBACK, SAVEPOINT.
Page 3 of 71
 ▪ Purpose: Manage transactions within a database.
• Examples:
o CREATE TABLE students (id INT, name VARCHAR(50));
o INSERT INTO students (id, name) VALUES (1, 'John');
Q5) What are the core components and characteristics of an RDBMS?
Answer:
• Core Components:
o Tables (Relations): Data organized in rows and columns.
o Schemas: Logical structure defining the tables and their
relationships.
o Keys: Primary, Candidate, Foreign, and Composite keys used to
maintain data integrity and establish relationships.
o Indexes: Speed up data retrieval.
• Characteristics:
o Data Integrity: Ensures accuracy and consistency.
o Data Abstraction Levels: Physical, logical, and view.
o Relationships: Defined through primary and foreign keys.
o SQL: Standard language for querying and managing relational
databases.
Q6) What are different types of SQL operations used in databases, and how
do they function?
Answer:
• DDL (Data Definition Language):
o Used to define or modify database structures.
o Commands:
▪ CREATE: Creates database objects.
▪ ALTER: Modifies existing structures.
▪ DROP: Deletes database objects.
▪ Example:
Page 4 of 71
CREATE TABLE Students (ID INT PRIMARY KEY, Name VARCHAR(50));
➢ DML (Data Manipulation Language):
• Used to manipulate data within database objects.
• Commands: SELECT, INSERT, UPDATE, DELETE.
• Example:
INSERT INTO Students (ID, Name) VALUES (1, 'Alice');
• DCL (Data Control Language):
o Manages permissions and user access.
o Commands: GRANT, REVOKE.
• TCL (Transaction Control Language):
o Manages database transactions.
o Commands: COMMIT, ROLLBACK, SAVEPOINT.
Q7) How are constraints used in databases to enforce data integrity?
Answer:
• Definition: Constraints are rules applied to database columns to ensure
data accuracy and consistency.
• Types of Constraints:
o Primary Key: Uniquely identifies each record.
o Foreign Key: Maintains referential integrity between tables.
o Unique: Ensures all values in a column are distinct.
o Not Null: Ensures a column cannot have a NULL value.
o Check: Limits the values that can be placed in a column.
o Example:
CREATE TABLE Employees (
EmpID INT PRIMARY KEY,
Name VARCHAR(50) NOT NULL,
Age INT CHECK (Age >= 18)
);
Page 5 of 71
➢ MODULE 2
Q1) What are the types of database systems?
Answer:
• Centralized Database Systems:
o All data is maintained at a single site.
o Processing of individual transactions is sequential.
o Common in organizations for localized data management.
• Parallel Databases:
o Improves processing and input/output speeds using multiple CPUs
and disks.
o Operations like loading data, building indexes, and evaluating
queries are parallelized.
o Parallel processing can occur on a single machine or across multiple
machines.
• Distributed Databases:
o Data is stored across multiple sites, and each site runs independently.
o Appears as a single database to users but is spread across various
nodes connected via networks.
Q2) What is parallel processing and why is it essential?
Answer:
• Definition:
o Parallel processing refers to the simultaneous execution of multiple
processes to enhance performance.
• Types of Parallel Systems:
o Coarse-grain: Small number of powerful processors.
o Fine-grain: Thousands of smaller processors.
• Key Performance Goals:
o Throughput: Number of tasks completed in a time interval.
o Response Time: Time to complete a single task.

Page 6 of 71
Q3) What is Speedup and Scaleup in parallel databases?
Answer:
• Speedup:
o Decreasing task execution time by increasing parallelism.
o Formula: Speedup = Time on Single Machine / Time on Parallel
System.
o Linear speedup is ideal where performance improves directly with
added resources.
• Scaleup:
o Maintaining performance levels as workload and resources increase
proportionally.
o Formula: Scaleup = Small System Elapsed Time / Large System
Elapsed Time.
• Factors Limiting Speedup and Scaleup:
o Startup Costs: Overhead in initiating multiple processes.
o Interference: Competition for shared resources like memory and
disks.
o Skew: Variance in execution times across parallel tasks.
Q4) What are the physical architectures for parallel databases?
Answer:

Page 7 of 71
Shared Memory:
• Description:
o Multiple CPUs access a common memory via an interconnection
network.
• Advantages:
o Efficient data access and inter-processor communication.
• Disadvantages:
o Increased waiting time due to contention.
o Scalability issues due to bottlenecks in memory access.
Shared Disk:
• Description:
o CPUs have private memory but access all disks via a network.
• Advantages:
o Fault tolerance: Tasks can be redistributed if a processor fails.
• Disadvantages:
o Limited scalability as the interconnection channel becomes a
bottleneck.
Shared Nothing:
• Description:
o Processors have their own memory and disks, communicating via
high-speed networks.
• Advantages:
o High scalability and minimal interference.
• Disadvantages:
o High cost of communication for non-local data access.
Hierarchical Systems:
• Combines shared-memory, shared-disk, and shared-nothing architectures.

Page 8 of 71
Q5) What is Parallel Query Evaluation?
Answer:
• Concept:
o Executing multiple queries or parts of a single query simultaneously.
o Emphasis on parallel execution of a single query.
• Techniques:
o Pipelined Parallelism: Operators producing and consuming data
simultaneously.
o Independent Parallelism: Operators execute independently.
o Data-Partitioned Parallelism: Input data is partitioned and
processed in parallel.

Q6) What are the types of data partitioning in parallel databases?

Answer:
Round-Robin Partitioning:

Page 9 of 71
 • Records are allocated sequentially to processors.
• Ensures even load distribution but lacks semantic grouping.
• Ideal for sequential scans.
Hash Partitioning:
• Data is grouped based on a hash function.
• Supports point queries efficiently.
Range Partitioning:
• Data is divided based on value ranges.
• Supports point and range queries.
Comparison of Partitioning Techniques:
• Round-Robin is efficient for full data scans.
• Hash and Range partitioning are better for subset queries.

Q7) What are distributed databases?

Answer:
• Distributed database system is a system where data is stored across several
sites, and each site is typically managed by a DBMS that can run
independent of the other sites.
• A distributed database appears to a user as a single database but is, in fact,
a set of databases stored on multiple computers.
• Computers in such systems are called sites or nodes.
• Nodes are connected through some communication medium like high-
speed network or telephone lines.

Page 10 of 71
 • A distributed database management system (DDBMS) is a centralized
software system that manages a distributed database in a manner as if it
were all stored in a single location.

Q8) What are the features, advantages, and disadvantages of distributed

databases?
Answer:
Features:
• Data is stored across multiple sites but appears as a single database to users.
• Supports data sharing, high availability, and local autonomy.
Advantages:
• Scalability: Easy to add new nodes and expand.
• Improved Reliability: Continues operation even with site failures.
• Lower Communication Costs: Local data handling reduces transfer.
Disadvantages:
• Complex software requirements.
• High processing overhead for maintaining consistency.
• Data integrity challenges.
Q9) What are Homogeneous Distributed Databases? Explain in detail.
Answer:
Homogeneous distributed databases are databases where all the sites (nodes)
use the same database management system (DBMS) software and have a unified
schema. These databases are typically easier to design and manage due to their
uniformity. Below are the detailed characteristics and advantages:

Page 11 of 71
Key Features:
• Same DBMS Software: All nodes in the system use the same DBMS,
which ensures compatibility and consistency.
• Uniform Schema: The database schema is the same across all sites,
simplifying database operations.
• Easy Communication: Nodes can easily communicate with one another
due to the unified system structure.
• Centralized Control: Often managed under a central administrative
system, making operations more streamlined.
• Data Distribution: Data is distributed across nodes in a way that maintains
consistency and reliability.
Advantages:
• Simplified Management: Easier to manage because of the consistent
environment.
• Efficient Query Processing: Uniformity allows for optimized query
execution across nodes.
• Reduced Complexity: Less complexity in handling database operations
and system integration.
• Improved Reliability: Easier to implement fault-tolerance mechanisms
due to the standardized architecture.
Example: A chain of retail stores using a single database system, such as Oracle
DBMS, across all its locations to manage inventory and sales data.

Q10) What are Heterogeneous Distributed Databases? Explain in detail.

Answer:

Page 12 of 71
Heterogeneous distributed databases are databases where the sites (nodes) may
use different DBMS software and may have different schemas. These systems are
more complex because they require additional layers for integration and
communication.
Key Features:
• Different DBMS Software: Nodes may use different types of database
management systems (e.g., Oracle, MySQL, SQL Server).
• Diverse Schemas: The schema at each site can vary, necessitating schema
mapping or transformation.
• Middleware Requirement: Middleware is required to enable
communication and data exchange between different nodes.
• Autonomous Nodes: Each node can operate independently, often with its
own administrative controls.
• Data Distribution: Data can be distributed in diverse formats and storage
structures.
Advantages:
• Flexibility: Can integrate various types of databases, making it suitable for
organizations with diverse systems.
• Scalability: New systems can be added without needing to standardize the
existing ones.
• Autonomy: Nodes can function independently, ensuring localized control
and management.
Challenges:
• Complex Integration: Requires extensive middleware and mapping
techniques to handle compatibility issues.
• Performance Overhead: Additional layers for communication and
translation can impact performance.
• Data Consistency: Ensuring consistency across different systems can be
challenging.
Example:

Page 13 of 71
 • A multinational corporation with multiple subsidiaries using different
database systems (e.g., one uses Oracle, another uses MySQL) but needing
to integrate their data for centralized reporting.

Q11) What are fragmentation and replication in distributed databases?

Answer:
Fragmentation:
• Dividing data into smaller parts stored across sites.
• Types:
o Horizontal Fragmentation: Splits data rows.

o Vertical Fragmentation: Splits data columns.

Page 14 of 71
 o Mixed Fragmentation: Combines horizontal and vertical
fragmentation.
• Correctness Criteria:
o Completeness: All data must be in some fragment.
o Reconstruction: Fragments can be combined to recreate the original
dataset.
o Disjointness: Data must not overlap between fragments.
Replication:
• Copies of data are stored at multiple sites.
• Types:
o Full Replication: Entire data is stored at all sites.
o No Replication: Data is stored only at one site.
o Partial Replication: Selective data is replicated at specific sites.
• Advantages:
o Improved availability and faster query processing.
• Disadvantages:
o Increased update costs and storage requirements.
Q12) What are the architectures of DDBMS?
Answer:
Client-Server Systems:
• Clients handle user interactions, and servers manage data and transactions.
• Advantages:
o Centralized control and efficient server utilization.
• Disadvantages:
o Does not support queries spanning multiple servers.

Page 15 of 71
Collaborating Server Systems:
• Servers work together to execute queries spanning multiple databases.
• Advantages:
o Efficient execution of complex queries.

Middleware Systems:
• Middleware coordinates query execution across servers.
• Useful for integrating legacy systems.

Q13) Distinguish between Distributed and Parallel Systems?

Answer:

Page 16 of 71
Page 17 of 71
➢ MODULE 3
Q1) OLAP vs OLTP
Answer:
OLAP (Online Analytical Processing) and OLTP (Online Transaction Processing)
are two types of systems designed to handle different kinds of data operations:
• OLAP:
o Focuses on complex queries and analytics, used primarily for decision
support and business intelligence.
o Deals with historical data and large volumes of data, optimized for read-
heavy operations.
o Examples: Data mining, trend analysis, reporting.
• OLTP:
o Focuses on real-time transaction processing, used in day-to-day operations.
o Deals with current data and frequent insert/update/delete operations.
o Examples: Bank transactions, order processing, customer data
management.

Key Differences:
- Purpose: OLAP is for analysis and decision-making, OLTP is for day-to-day
transaction processing.
- Data Volume: OLAP handles large datasets, OLTP handles smaller datasets
with frequent updates.
- Operations: OLAP supports complex queries, OLTP handles fast, simple
queries (often using indexes).

Page 18 of 71
Q2) Advantages of OLAP
Answer:
OLAP provides several benefits, especially in data analysis and business
intelligence:
• Fast Query Processing: OLAP allows quick retrieval of summarized data
and complex calculations.
• Multidimensional Analysis: Users can view data from multiple
dimensions (e.g., time, geography, product category).
• Data Consolidation: OLAP systems can combine data from various
sources, making it easier to analyze large amounts of data.
• Decision Support: Provides support for business decision-making by
offering insights based on historical data trends.
• What-If Analysis: Users can model different scenarios to see potential
outcomes.

Page 19 of 71
Q3) OLAP Operations
Answer:
OLAP supports several key operations for multidimensional analysis:
- Slice: Selecting a single layer from a multidimensional dataset (e.g., viewing
data for a specific time period).
- Dice: Selecting a sub-cube (a set of values that fit specific criteria, e.g., viewing
sales data for a particular region and year).
- Drill-Down: Zooming into more detailed data (e.g., from yearly sales to
monthly or daily sales).
- Roll-Up: Aggregating data (e.g., summarizing daily data into monthly data).
- Pivot: Re-orienting the data cube to view it from different perspectives
Q4) ROLAP vs MOLAP
Answer:

Q5) Advantages & Disadvantages of MOLAP and ROLAP.

Answer:

Page 20 of 71
• MOLAP:
▪ Advantages:
- Fast query performance due to pre-aggregation.
- Optimized for read-heavy workloads and complex queries.
- Compact storage as data is often compressed.
▪ Disadvantages:
- Limited scalability with large datasets.
- Inflexible with regard to data changes (since data is pre-aggregated).
- Higher costs for data storage and maintenance.
• ROLAP:
▪ Advantages:
- More scalable for very large datasets.
- Flexible; queries can be more dynamic as data is not pre-aggregated.
- Uses standard relational databases which are widely available and familiar
to developers.
▪ Disadvantages:
- Slower query performance compared to MOLAP because it computes
aggregations in real-time.
- Requires more complex and optimized queries.

Q6) Data Warehouse in Detail, Advantages, and Characteristics.

Answer:
A Data Warehouse (DW)is a central repository designed for reporting and
analysis of data. It consolidates data from various sources to support decision-
making and business intelligence.
▪ Characteristics:
- Subject-Oriented: Focuses on key business areas like sales, finance, or
marketing.
- Integrated: Combines data from different sources into a consistent format.
- Time-Variant: Data is stored over time, enabling historical analysis.

Page 21 of 71
 - Non-Volatile: Once data is entered, it is not frequently changed, allowing
for stable historical analysis.
▪ Advantages:
- Consolidated View: Provides a unified view of data from multiple sources.
- Business Intelligence: Supports advanced analytics and reporting for
decision-making.
- Improved Data Quality: ETL processes can clean and transform data
before loading it into the warehouse.
- Historical Data Analysis: Facilitates trend analysis and forecasting by
storing time-series data.

Q7) Database vs Data Warehouse

Answer:

Key Differences:
- Purpose: Databases focus on operational tasks, data warehouses focus on
analytical tasks.

Page 22 of 71
- Data Volume: Data warehouses handle much larger volumes of data.
- Data Processing: Databases handle real-time transactions, data warehouses
handle historical queries.
Q8) Data Mart in Detail with Advantages and Disadvantages.
Answer:
A Data Mart is a subset of a data warehouse focused on a specific business area
or department (e.g., marketing, sales, finance).
▪ Advantages:
- Faster Query Performance: Since data marts focus on specific areas,
queries are typically faster.
- Cost-Effective: Smaller in scope and size, so they can be less expensive to
build and maintain.
- Simplified Design: Easier to design and implement than a full-scale data
warehouse.
▪ Disadvantages:
- Limited Scope: Data marts only focus on specific business areas, which
can limit cross-departmental insights.
- Data Silos: They can create silos of data if not properly integrated with the
larger data warehouse.
- Data Redundancy: Multiple data marts may duplicate data from the central
warehouse.

Q9) Data Warehouse vs Data Mart.

Answer:

Page 23 of 71
Key Differences:
- Scope: Data warehouses serve the entire organization, while data marts serve
specific departments.
- Cost and Complexity: Data marts are less expensive and easier to implement
than data warehouses.
- Data Integration: Data warehouses are integrated across multiple systems,
while data marts may only pull from a subset.
Q10) Data Warehouse Design Approaches (Top-Down and Bottom-Up).
Answer:
There are two main approaches to designing a data warehouse:
• Top-Down Approach:
- The data warehouse is designed first as a central repository.
- Data marts are created as needed, drawing data from the central warehouse.
- Example: Bill Inmon’s approach.
- Advantages: Provides a comprehensive view of enterprise data and reduces
- redundancy.
- Disadvantages: More time-consuming and expensive to implement.

Page 24 of 71
• Bottom-Up Approach:
- Starts with building individual data marts, which can be integrated later into
a larger data warehouse.
- Example: Ralph Kimball’s approach.
- Advantages: Faster implementation and allows for early value from data
marts.
- Disadvantages: May result in data silos and redundancy.

Q11) What is the ETL process in the context of Data Warehousing?

Answer:
The ETL process stands for Extract, Transform, and Load. It is a critical process
in Data Warehousing and comprises the following stages:
• Extract:
o Collects data from various operational systems, such as OLTP
databases, or external sources like files or APIs.
o Can involve static extraction (a one-time snapshot) or incremental
extraction (capturing changes since the last extract).
o Ensures the selection of relevant data for loading into the warehouse.
• Transform:
o Converts data into a format suitable for analysis in the data
warehouse.
o Includes operations like:
▪ Data cleansing (removing errors like misspellings or duplicate
data).
▪ Data reformatting (standardizing formats).
▪ Key generation (creating unique identifiers).
▪ Applying business rules to ensure consistency.
▪ Aggregation (summarizing data, e.g., monthly sales totals).
• Load:
o Transfers the transformed data into the data warehouse.
o Can occur in two modes:

Page 25 of 71
 ▪ Refresh Mode: Bulk rewriting of the data periodically.
▪ Update Mode: Writing only incremental changes since the
last load.
o Builds aggregates and creates indexes for optimized querying.
Q12) Why is the ETL process considered time-consuming in Data
Warehousing?
Answer:
The ETL process is time-consuming because:
• It involves extensive data cleansing to resolve issues like inconsistencies,
missing data, and errors.
• Requires complex transformations to integrate data from heterogeneous
sources.
• Demands significant resources for testing, debugging, and ensuring
accuracy.
• Approximately 80% of the development time in a data warehouse project
is spent on ETL tasks due to its intricacy.
Q13) Explain the cleansing step in the ETL process.
Answer:
Data cleansing is a critical part of the transformation step in ETL and includes the
following tasks:
• Correcting misspellings and erroneous entries.
• Standardizing inconsistent formats (e.g., date formats).
• Removing duplicate records to prevent redundancy.
• Filling in missing values to ensure completeness.
• Verifying data integrity (e.g., ensuring references match between tables).
• Implementing business rules to validate data accuracy.
Advanced methods like pattern recognition and AI techniques are often
employed to enhance data quality during cleansing.
Q14) What is dimensional modeling, and how does it differ from ER
modelling?
Answer:
Dimensional modeling is a design technique used in data warehousing that
focuses on structuring data for analysis.

Page 26 of 71
 • Key Characteristics:
o Focuses on measures (facts) and business dimensions.
o Designed for query optimization and user understandability.
o Encourages denormalization for better performance.
Differences between Dimensional and ER Modeling:

Q15) What are the steps of Dimensional Modeling?

Answer:
Dimensional modeling involves structuring data around business processes and
decision-support requirements. The steps are as follows:
1. Choose the Process:
o Identify the business process for analysis (e.g., sales, inventory
management).
o Define the subject areas, such as customer behavior or product
performance.
2. Choose the Grain:
o Determine the level of detail (granularity) for the data.
o Example: In a sales process, the grain could be daily sales at the store
level.
3. Identify and Confirm Dimensions:
o Select dimensions that describe the facts.
o Examples: Product, Time, Customer, and Location.
o Verify that these dimensions meet analytical requirements.
4. Identify the Facts:

Page 27 of 71
 o Select measurable metrics (facts) like sales amount, units sold, or
profit.
o Ensure these facts align with business needs.
5. Choose the Duration:
o Decide how much historical data to store in the warehouse.
o Consider factors like compliance, trend analysis, and storage
capacity.

Q16) What is a Fact Table?

Answer:
A fact table stores quantitative data (facts) related to business processes.
• Key Features:
o Contains measurements or metrics for analysis.
o Includes foreign keys referencing dimension tables.
o Stores data at a detailed or aggregated level.
o Examples of facts: Revenue, Quantity Sold, and Profit.
Q17) What is a Dimension Table?
Answer:
A dimension table stores descriptive attributes related to facts, providing context
for analysis.
Page 28 of 71
 • Key Features:
o Contains textual, descriptive values (e.g., Customer Name, Product
Category).
o Includes attributes that allow users to filter or group data.
o Dimension tables are usually wide with many columns.
o Connected to the fact table via foreign keys.
Q18) What are the differences between Fact and Dimension Tables?
Answer:

Q19) Explain the Star Schema in detail, including Dimension Table and Fact
Table.
Answer:
A Star Schema is the simplest type of data warehouse schema, consisting of one
central fact table surrounded by dimension tables.
• Fact Table:
o Stores quantitative metrics (facts).
o Primary key: A concatenation of foreign keys from dimension tables.
o Data is stored at the lowest level of granularity.
• Dimension Table:
o Stores attributes that describe facts.
o Example attributes: Product Name, Time Period, Customer
Demographics.
o Dimension tables are de-normalized for better query performance.
• Example Structure:
Page 29 of 71
 o Fact Table: Sales (Sales Amount, Units Sold).
o Dimensions: Product, Customer, Time, Store.
Q20) What are the characteristics of a Star Schema?
Answer:
• De-normalized Structure: Simplifies queries by reducing joins.
• One-to-Many Relationship: Fact table is at the center, linked to multiple
dimension tables.
• Optimized for Query Performance: Faster aggregations and drill-downs.
• Supports Multidimensional Analysis: Users can analyze data across
various dimensions (e.g., sales by region and time).
• Easy to Understand: Clear relationships between facts and dimensions.
Q21) Explain the Snowflake Schema in detail.
Answer:
The Snowflake Schema is an extension of the Star Schema where dimension
tables are normalized into multiple related tables.
• Key Characteristics:
o Dimension tables are divided into sub-tables to eliminate
redundancy.
o Example: A Product Dimension may have separate tables for
Category, Brand, and Product Details.
o Requires multiple joins to retrieve data.
• Advantages:
o Saves storage space by eliminating redundant data.
o Easier to update and maintain normalized structures.
• Disadvantages:
o Complex structure, harder for users to understand.
o Slower query performance due to multiple joins.
Q22) How does the Star Schema differ from the Snowflake Schema?
Answer:

Page 30 of 71
Q23) What is Metadata, and Why is it Important?
Answer:
Metadata is often described as "data about data." It provides information about
the structure, organization, and content of data in a data warehouse, enabling
users and systems to understand, manage, and use the data effectively.
Key Components of Metadata:
• Data Source Information:
o Specifies the origin of the data, such as databases, files, or APIs.
o Tracks changes in source systems over time.
• Data Transformation Rules:
o Includes business logic applied during the ETL process.
o Defines how raw data is converted into meaningful information.
• Relationships and Hierarchies:
o Explains how tables, dimensions, and facts are interconnected.
o Documents hierarchies (e.g., Year → Quarter → Month → Day).
• User Navigation:
o Helps users locate and access relevant data easily.
o Provides context to simplify data exploration.
Importance:
• Acts as a directory for locating data in the warehouse.
• Helps developers maintain the data warehouse.
• Enables end-users to understand data in business terms.

Page 31 of 71
Q24) Explain Operational Metadata in detail (Types of Metadata).
Answer:
Operational metadata describes the source systems, the structure of the data, and
how the data is extracted from operational systems into the data warehouse.
• Key Features:
o Provides detailed information about source systems, such as data
structures, data types, and relationships.
o Captures the history and changes made to source data during its
journey to the data warehouse.
o Tracks record-level details, including:
▪ Field lengths and formats.
▪ Encoding and decoding schemes.
▪ Constraints such as primary keys and foreign keys.
• Examples of Operational Metadata:
o A table from the source system contains employee data:
▪ Employee ID (Primary Key).
▪ First Name (String, 50 characters).

Page 32 of 71
 ▪ Date of Joining (Date Format: DD-MM-YYYY).
o Describes relationships such as one-to-many or many-to-many in
the source data.
• Use Case:
o Helps trace back any piece of data in the data warehouse to its
original source and format.
Q25) Explain Extraction and Transformation Metadata in detail (Types of
Metadata).
Answer:
Extraction and transformation metadata focuses on the process of transferring
data from source systems to the data warehouse.
• Key Features:
o Tracks data extraction processes, including:
▪ Extraction methods (e.g., static or incremental).
▪ Extraction frequency (e.g., hourly, daily, weekly).
o Documents data transformation rules, such as:
▪ Conversions (e.g., currency exchange or unit conversions).
▪ Aggregations (e.g., calculating total sales for a region).
▪ Joins and merges between datasets from multiple sources.
o Provides logs of errors and corrections during transformation.
• Examples of Extraction and Transformation Metadata:
o Business rules applied during transformation:
▪ Combine "First Name" and "Last Name" into a single field
"Full Name."
▪ Convert all dates into a consistent format (e.g., YYYY-MM-
DD).
o A transformation rule that aggregates daily sales into monthly totals.
• Use Case:
o Ensures that the data transformation adheres to business logic and
helps troubleshoot any issues during the ETL process.
Page 33 of 71
Q26) Explain End-User Metadata in detail (Types of Metadata).
Answer:
End-user metadata is designed to make the data warehouse accessible and
understandable to end-users, allowing them to navigate the data warehouse
intuitively.
• Key Features:
o Provides a business-friendly layer that translates technical terms
into meaningful business terminology.
o Acts as a navigational map, showing the data's structure and
hierarchy.
o Describes how to query and report on data effectively.
o Includes user-defined views, filters, and aggregations for ease of
use.
• Examples of End-User Metadata:
o A sales executive sees terms like "Product Category" and "Monthly
Revenue" instead of technical table names like prod_cat or
rev_month.
o Hierarchies defined for drill-downs:
▪ Time Dimension: Year → Quarter → Month → Day.
▪ Geography Dimension: Country → State → City → Store.
• Use Case:
o Helps non-technical users interact with the data warehouse to
generate reports, visualize trends, and make decisions.

Page 34 of 71
➢ MODULE 4
Q1) What is data mining, and what are its primary functionalities?
Answer:
• Definition of Data Mining:
o The process of discovering implicit, previously unknown, and
potentially useful information from large datasets.
o A key component of the Knowledge Discovery in Databases (KDD)
process.
• Primary Functionalities of Data Mining:
o Classification:
▪ Predicts categorical class labels by constructing a model
based on a training dataset.
o Regression:
▪ Maps data to a real-valued prediction variable.
o Clustering:
▪ Identifies a finite set of categories or clusters to describe the
data.
▪ Groups similar data points and separates dissimilar ones.
o Dependency Modeling:
▪ Describes significant dependencies between variables in a
dataset.
o Deviation and Change Detection:
▪ Identifies significant changes or anomalies in the data.
o Summarization:
▪ Provides a compact description for a subset of data.
Q2) What are the steps in the Knowledge Discovery in Databases (KDD)
process?
Answer:
1. Data Cleaning:

Page 35 of 71
 Removes missing, noisy, and inconsistent data.
Essential for preparing quality data for analysis.
2. Data Integration:
Combines heterogeneous data from multiple sources.
Creates a unified and consistent dataset.
3. Data Selection:
Extracts data relevant to the analysis from a larger dataset.
4. Data Transformation:
Converts data into suitable formats for mining.
Includes summary operations like aggregation and normalization.
5. Data Mining:
Applies intelligent methods or models to extract useful data patterns.
6. Pattern Evaluation:
Identifies interesting and useful patterns based on specific measures.
7. Knowledge Representation:
Utilizes visualization and representation techniques to present mining
results effectively.

Q3) What is the significance of data preprocessing in data mining?

Answer:
• Purpose:
Page 36 of 71
 o Ensures the dataset's information content is maximized for mining
tools.
• Key Reasons for Importance:
o Quality Decisions Depend on Quality Data:
▪ Incomplete or duplicate data can lead to incorrect or
misleading conclusions.
o Data Consistency:
▪ Ensures uniform data integration in a data warehouse.
o Majority of Effort in Data Warehouse Building:
▪ Data extraction, cleaning, and transformation comprise the
bulk of the work.
o Data Formatting:
▪ Prepares data to be compatible with specific software tools
and methods.
Q4) What are the key steps involved in data preprocessing?
Answer:
• 1. Data Cleaning:
o Addresses missing values, noisy data, and inconsistencies.
o Methods include:
▪ Filling missing values.
▪ Smoothing noisy data.
▪ Identifying or removing outliers.
▪ Resolving inconsistencies.
• 2. Data Integration:
o Combines data from multiple sources to provide a unified view.
o Detects and resolves data conflicts and redundancy.
• 3. Data Transformation:
o Converts data into forms suitable for mining.
o Techniques include:
Page 37 of 71
 ▪ Normalization.
▪ Aggregation.
▪ Discretization and concept hierarchy generation.
• 4. Data Reduction:
o Reduces the volume of data while maintaining its analytical utility.
o Methods include:
▪ Data cube aggregation.
▪ Attribute subset selection.
▪ Dimensionality reduction.
▪ Sampling and discretization.
Q5) What is data cleaning, and what are its key tasks?
Answer:
• Importance:
o "Data cleaning is one of the three biggest problems in data
warehousing" — Ralph Kimball.
o It is essential to ensure the data's completeness, accuracy, and
consistency.
• Tasks:
o Filling in missing values.
o Identifying and smoothing noisy data.
o Correcting inconsistent data.
o Resolving redundancy caused by data integration.
• Challenges with Real-World Data:
o Incomplete Data:
▪ Missing attribute values or aggregate data.
o Noisy Data:
▪ Contains errors or outliers (e.g., incorrect salary or age
values).

Page 38 of 71
 o Inconsistent Data:
▪ Discrepancies in codes or naming conventions across
datasets.
Q6) How can missing data and noisy data be handled?
Answer:
• Handling Missing Data:
o Ignoring tuples with missing values (not effective for high
percentages of missing data).
o Filling missing values:
▪ Manually: Tedious and infeasible for large datasets.
▪ Automatically:
▪ Using a global constant (e.g., "unknown").
▪ Attribute mean, median, or mode.
▪ Most probable value based on inference techniques
(e.g., Bayesian formula, decision tree).
• Handling Noisy Data:
o Binning:
▪ Sorting data into bins and smoothing by means, medians, or
boundaries.
o Regression:
▪ Fitting data into regression functions for smoothing.
o Clustering:
▪ Identifying and removing outliers using clustering algorithms.
o Combined Computer and Human Inspection:
▪ Detecting and validating suspicious values manually.
Q7) What is data integration, and how are redundancy and conflicts
handled?
Answer:
• Definition:

Page 39 of 71
 o Combines data from multiple sources into a coherent and unified
dataset.
• Challenges:
o Data Value Conflicts:
▪ Differences in representation, scales (e.g., metric vs. British
units), or entity naming.
o Entity Identification Problem:
▪ Schema integration and object matching to identify real-world
entities across datasets.
• Handling Redundancy:
o Redundant data arises during database integration.
o Detection Methods:
▪ Correlation analysis to identify redundant attributes.
▪ Checking for derived attributes that repeat existing
information (e.g., annual revenue derived from monthly
revenue).
Q8) What is data transformation, and what are its techniques?
Answer:
• Definition:
o Data transformation involves converting data into formats suitable
for data mining.
o This ensures the data is in a consistent, meaningful structure for
analysis.
• Key Techniques:
o Smoothing:
▪ Reducing noise in data using binning, regression, or clustering
methods.
o Aggregation:
▪ Summarizing data by applying operations like averages,
sums, or counts (e.g., daily sales aggregated to monthly
totals).
Page 40 of 71
 o Normalization:
▪ Scaling data to fit within a specific range (e.g., -1.0 to 1.0 or
0.0 to 1.0).
▪ Example: Min-max normalization.
o Discretization:
▪ Replacing continuous attribute values with interval labels or
categories (e.g., dividing ages into "below 18," "18-30," and
so on).
▪ Techniques include binning, histogram analysis, and decision
trees.
o Concept Hierarchy Generation:
▪ Converting lower-level data (e.g., city names) into higher-
level data (e.g., countries) for easier analysis.
Q9) What is data reduction, and why is it important?
Answer:
• Definition:
o Data reduction reduces the volume of data while retaining its
analytical integrity.
• Importance:
o Helps manage and analyze large datasets efficiently.
o Saves computational time and storage.
o Essential for terabytes of data in data warehouses.
• Techniques:
o Data Cube Aggregation:
▪ Applies aggregation at multiple levels, reducing data
complexity.
o Dimensionality Reduction:
▪ Reduces attributes using feature selection or variable
composition.
o Numerosity Reduction:

Page 41 of 71
 ▪ Stores models of data instead of the entire dataset (e.g.,
histograms, clustering, or regression).
o Data Compression:
▪ Uses encoding to reduce dataset size.
o Discretization and Concept Hierarchy Generation:
▪ Converts continuous data to categorical data.
Q10) What are the methods for handling redundancy in data integration?
Answer:
• Handling Redundancy:
o Derived Attributes:
▪ Detect attributes derived from others (e.g., annual revenue
derived from monthly revenue).
o Inconsistencies in Naming:
▪ Resolve issues caused by differences in attribute names across
databases.
o Correlation Analysis:
▪ Identifies redundancy by measuring attribute correlations.
o Schema Integration:
▪ Combines schemas to eliminate duplicated structures.
Q11) What are the major techniques of dimensionality reduction?
Answer:
• Techniques:
o Feature Reduction:
▪ Combines less important features to create new, synthetic
ones using linear combinations.
▪ Example: Average miles driven = mileage / (current year –
year of purchase).
o Feature Selection:

Page 42 of 71
 ▪ Identifies a subset of variables or features that are most
relevant for the problem.
o Parametric Methods:
▪ Assumes the data fits a model (e.g., regression, log-linear
models) and only stores parameters.
o Non-Parametric Methods:
▪ Does not assume any specific data model.
▪ Example: Histograms, clustering, and sampling.
Q12) What is KDD (Knowledge Discovery in Databases), and what are its
steps?
Answer:
• Definition:
o KDD refers to extracting implicit, useful knowledge from data
stored in databases.
• Steps in KDD:
o Data Cleaning:
▪ Removing missing, noisy, or inconsistent data.
o Data Integration:
▪ Combining heterogeneous data from various sources.
o Data Selection:
▪ Retrieving relevant data for analysis.
o Data Transformation:
▪ Summarizing and converting data into suitable forms.
o Data Mining:
▪ Using intelligent models to uncover useful patterns.
o Pattern Evaluation:
▪ Evaluating the interestingness of discovered patterns.
o Knowledge Representation:
▪ Presenting results using visualization techniques.
Page 43 of 71
Q13) What are the applications of data mining?
Answer:
• Applications:
o Banking:
▪ Loan/credit card approval.
▪ Predicting customer behavior based on historical data.
o Customer Relationship Management:
▪ Identifying customers likely to switch to competitors.
o Targeted Marketing:
▪ Identifying potential responders to promotions.
o Fraud Detection:
▪ Detecting fraudulent activities in telecommunications and
financial transactions.
o Manufacturing:
▪ Automatically adjusting process parameters.
o Medicine:
▪ Predicting disease outcomes and treatment effectiveness.
o Scientific Analysis:
▪ Identifying new galaxies or scientific discoveries through
clustering.

Page 44 of 71
Q14) What are the differences between supervised and unsupervised
learning?
Answer:

Q15) What are the types of sampling in data reduction?

Answer:
• Simple Random Sampling:
o Each data point has an equal probability of selection.
• Simple Random Sampling Without Replacement (SRSWOR):
o Once selected, an object is removed from the population.
• Simple Random Sampling With Replacement (SRSWR):

Page 45 of 71
 o Selected objects remain in the population for possible reselection.
• Stratified Sampling:
o Divides the dataset into partitions and samples proportionally from
each partition.
Q16) Database Processing vs Data Mining Processing.
Answer:

Q17) Explain data mining techniques in detail.

Answer:
Data mining techniques are broadly classified into two categories: Descriptive
and Predictive. Each serves specific purposes in extracting useful patterns and
insights from data.
1. Descriptive Techniques
These techniques are used to summarize or describe data. They help identify
patterns and relationships within a dataset.
• Clustering:
o Groups data points into clusters based on similarity.
o Ensures that objects in the same cluster are similar and those in
different clusters are dissimilar.

Page 46 of 71
 o Example: Segmenting customers into groups based on purchasing
behavior.
• Association:
o Identifies relationships between variables in a dataset.
o Example: Discovering items frequently bought together, like "bread
and butter."
• Summarization:
o Provides a compact representation of the data.
o Example: Generating summary statistics for different customer
groups.
2. Predictive Techniques
These techniques are used to predict unknown values or future outcomes based
on current data.
• Classification:
o Assigns items to predefined categories or classes.
o Example: Predicting whether a customer will default on a loan
(Yes/No).
• Regression:
o Maps a data item to a real-valued variable, such as predicting sales
for the next quarter.
• Sequential Analysis:
o Discovers sequential patterns in the data.
o Example: Identifying customer purchase sequences (e.g., customers
who buy a phone may buy a phone case next).
• Decision Tree:
o Uses a tree-like model for decisions and their possible
consequences.
o Example: Determining loan approval based on income, age, and
credit score.
• Rule Induction:

Page 47 of 71
 o Generates rules from the data to explain patterns.
o Example: "If a customer buys Product A, they are 80% likely to buy
Product B."
• Neural Networks:
o Mimics the human brain to recognize patterns and learn from data.
o Example: Image recognition or detecting fraud in financial
transactions.
• Nearest Neighbour Classification:
o Classifies data points based on the closest data points in the training
dataset.
o Example: Recommending movies based on a user’s similarity to
others.

Page 48 of 71
➢ MODULE 5
Q1) What is frequent pattern analysis?
Answer:
• Definition:
o A frequent pattern is a set of items, subsequences, or substructures
that occur frequently in a dataset. It reflects repetitive or inherent
regularities in data, which can be leveraged for insightful analysis.
• Motivation:
o Frequent pattern analysis seeks to uncover patterns that provide
value in decision-making processes.
o Examples:
▪ Identifying products frequently bought together (e.g., bread
and butter) to design promotional strategies.
▪ Predicting subsequent purchases after a significant purchase
like a computer.
▪ Automatically categorizing web documents based on shared
patterns.
• Applications:
o Market Basket Analysis: Understand purchase behavior by
analyzing itemsets.
o Cross-marketing: Recommend related products based on frequent
co-purchases.
o Catalog Design: Optimize item placement for better customer
experience.
o Web Log Analysis: Track user navigation patterns.
o DNA Sequence Analysis: Identify repetitive biological patterns for
genetic research.
Q2) What is Association Rule Mining?
Answer:
• Definition:

Page 49 of 71
 o Association Rule Mining identifies interesting associations and
relationships among large sets of data items. It is a method for
discovering patterns, rules, or correlations within transactional
datasets.
• Key Components:
o Itemsets: Collections of items frequently found together in
transactions.
o Association Rules: Implication rules in the form , where and are
itemsets.
o Example: "Bread → Milk" implies that customers buying bread are
likely to buy milk.
• Applications:
o Market Basket Analysis: Analyze purchasing patterns.
o Webpage Recommendation: Identify correlations between
webpage visits.
o Customer Segmentation: Group customers with similar purchasing
behaviors.
• Evaluation Metrics:
o Support: Indicates how frequently an itemset appears in the dataset.
o Confidence: Measures the strength of implication.
Q3) What is Market Basket Analysis?
Answer:
• Definition:
o Market Basket Analysis is a data analysis technique that helps
retailers understand customer purchase behaviors by identifying
which items are frequently purchased together.
• How It Works:
o Items are grouped into transactions (baskets).
o Frequent itemsets and their associations are derived using
techniques like Apriori or FP-Growth algorithms.
• Applications:

Page 50 of 71
 o Store Layout Optimization: Arrange items frequently bought
together near each other.
o Targeted Promotions: Send personalized offers for frequently co-
purchased items.
o Cross-Selling: Recommend complementary items (e.g., chips with
soda).
• Broader Applications:
o Medical Diagnosis: Discover relationships between symptoms and
diseases.
o Fraud Detection: Identify suspicious patterns in financial
transactions.
o Bioinformatics: Analyze protein sequences for similarities and
anomalies.
Q4) Define important terms in Association Rule Mining.
Answer:
• Itemset:
o A collection of one or more items in a dataset.
o Example: {Bread, Milk}.
• k-itemset:
o An itemset containing items.
o Example: {Bread, Milk, Diaper} is a 3-itemset.
• Support Count (σ):
o The frequency of occurrence of an itemset in a dataset.
o Example: σ({Milk, Bread, Diaper}) = 2 indicates that the itemset
appears in 2 transactions.
• Support (s):
o The proportion of transactions containing an itemset.
o Formula: , where is the total number of transactions.
• Frequent Itemset:

Page 51 of 71
 o An itemset with a support value greater than or equal to the user-
defined minimum support threshold (minsup).
Q5) What is the Apriori Algorithm?
Answer:
• Definition:
o Apriori is an iterative algorithm used to find frequent itemsets in a
dataset. It relies on the Apriori property, which states that if an
itemset is infrequent, all its supersets are also infrequent. This anti-
monotone property of support ensures efficiency by reducing the
number of candidate itemsets.
• Key Features:
o Uses prior knowledge of frequent itemsets.
o Employs a level-wise search approach where frequent -itemsets are
used to generate -itemsets.
• Steps:
1. Generate Frequent 1-itemsets:
▪ Scan the database to calculate the support of each individual
item.
▪ Discard items with support below the minsup threshold.
▪ Retain the frequent 1-itemsets (L1).
2. Candidate Generation:
▪ Use the frequent -itemsets to generate candidate -itemsets.
▪ Apply the Apriori property to prune candidates that have
infrequent subsets.
3. Support Counting:
▪ Scan the database to count the support of each candidate
itemset.
▪ Eliminate candidates that do not meet the minsup threshold.
4. Iteration:
▪ Repeat steps 2 and 3 until no new frequent itemsets are
generated.
Page 52 of 71
 5. Generate Association Rules:
▪ For each frequent itemset, generate all possible association
rules.
▪ Calculate the confidence for each rule and discard those below
the minconf threshold.
• Example:
o Input Data:
▪ Transactions: {a, b, c}, {a, c}, {a, d}, {b, e, f}.
▪ Minsup = 50%; Minconf = 50%.
o Iteration 1:
▪ Frequent 1-itemsets: {a}, {b}, {c}.
o Iteration 2:
▪ Generate candidate 2-itemsets: {a, b}, {a, c}, {b, c}.
▪ Frequent 2-itemset: {a, c}.
o Result:
▪ Frequent itemsets: {a, c}.
• Advantages:
o Reduces computation by pruning infrequent itemsets early.
o Handles large datasets effectively with iterative refinement.
• Limitations:
o Requires multiple scans of the database.
o Inefficient for datasets with high dimensionality or long itemsets.
Q6) Solve this: Determine association rules with 50% support and 75%
confidence for a dataset.
Answer:
• Example Data:
o Transactions:
▪ T1:{Card Reader, Mobile, Laptop}

Page 53 of 71
 ▪ T2:{Mobile, Laptop}
▪ T3:{Laptop, Camera}
▪ T4:{Card Reader, Laptop}
▪ T5:{Card Reader, Mobile}
• Analysis:
o Rule 1: Card Reader → Mobile
▪ Support: 35=60%
▪ Confidence: 34=75%
o Rule 2: Mobile → Laptop
▪ Support: 2/5=40%
▪ Confidence: 2/3=66.7%
Q7) What is Associative Classification (AC)?
Answer:
• Definition:
o A data mining technique that integrates association rule mining and
classification.
o Uses strong associations between frequent patterns and class labels
to build classifiers.
• Features:
o Focuses on generating Class Association Rules (CARs).
o Rules are of the form P1 ∧ P2∧⋯∧ Pk→C, where P1,P2,… are
attribute-value pairs, and C is a class label.
• Applications:
o Sentiment analysis (e.g., positive or negative review classification).
o Medical diagnosis (e.g., associating symptoms with diseases).
o Fraud detection (e.g., predicting fraudulent transactions).
Q8) What are the steps in Associative Classification?
Answer:

Page 54 of 71
 1. Generate Class Association Rules (CARs):
o Identify frequent itemsets with class labels.
o Only include itemsets that meet minimum support and confidence
thresholds.
2. Build a Classifier:
o Arrange CARs in descending order of confidence and support.
o Use the rules to predict class labels for new data.
3. Classify New Data:
o Apply the highest-ranking rule that matches the attributes of a data
instance.
Q9) What are Class Association Rules (CARs)?
Answer:
• Definition:
o A subset of association rules restricted to class labels on the right-
hand side (RHS).
• Characteristics:
o Frequent Rule Items: Rules with support above the minimum
support threshold.
o Accurate Rules: Rules with confidence above the minimum
confidence threshold.
• Example:
o Rule: {Bread, Milk}→Class: Purchased.
▪ Support = 40% (appears in 40% of transactions).
▪ Confidence = 80% (80% of transactions with Bread and Milk
are labelled as Purchased).
Q10) How is a classifier built using Associative Classification?
Answer:
• Procedure:
1. Rule Generation:

Page 55 of 71
 ▪ Generate all possible rules that meet the support and
confidence thresholds.
▪ Example:
▪ If (Bread AND Milk)→Purchased\text{If (Bread AND
Milk)} →
\text{Purchased}If (Bread AND Milk)→Purchased
(Support: 40%40\%40%, Confidence: 80%80\%80%).
2. Rule Ranking:
▪ Rank rules by:
▪ Confidence (higher is better).
▪ Support (secondary criterion if confidence is equal).
3. Classifier Construction:
▪ Use the ranked rules to build a classification model.
▪ For a new data instance, apply the first matching rule in the
list.
Q11) Solve this: Build an associative classifier for the following dataset.
Dataset:

Answer:
Steps:
1. Generate Frequent Itemsets:
o MinSupport = 25% (i.e., itemsets must appear in at least 2
transactions).
o Frequent itemsets:
▪ {Bread}: Appears in 4 transactions (Support = 80%).

Page 56 of 71
 ▪ {Milk}: Appears in 4 transactions (Support = 80%).
▪ {Diaper}: Appears in 3 transactions (Support = 60%).
▪ {Bread,Milk}: Appears in 3 transactions (Support = 60%).
▪ {Diaper,Beer}: Appears in 3 transactions (Support = 60%).
2. Generate Class Association Rules (CARs):
o Example rules:
▪ {Diaper, Beer}→Male (Support: 60%, Confidence: 100%).
▪ {Bread, Milk}→Female (Support: 40%, Confidence:66.7%).
3. Rank Rules:
o Rule 1: {Diaper, Beer}→Male (Confidence = 100%).
o Rule 2: {Bread, Milk}→Female (Confidence = 66.7%).
4. Build the Classifier:
o Rule-based classifier:
▪ If Diaper AND Beer, then Male\text{Male}Male.
▪ Else if Bread AND Milk, then Female.
Q12) What are the advantages of Associative Classification?
Answer:
• High Accuracy:
o Explores multiple attribute relationships simultaneously, yielding
better accuracy than single-attribute methods like decision trees.
• Interpretability:
o Rules are easy to understand and interpret (e.g., "If Diaper and Beer,
then Male").
• Flexibility:
o Can handle both numeric and categorical data after discretization.
• Effective Use of Associations:
o Finds hidden relationships in data.

Page 57 of 71
Q13) What are the limitations of Associative Classification?
Answer:
• Rule Explosion:
o Generates a large number of rules, making computation expensive.
• Overfitting:
o May create overly specific rules that perform poorly on new data.
• Dependency on Thresholds:
o Results heavily depend on minimum support and confidence
thresholds.
• Scalability:
o Computational cost increases significantly for large datasets.
Q14) What are common algorithms used in Associative Classification?
Answer:
• CBA (Classification by Association):
o Builds a classifier using Class Association Rules.
o Organizes rules by decreasing confidence and support.
• CMAR (Classification based on Multiple Association Rules):
o Uses statistical analysis of multiple rules for classification.
• CPAR (Classification based on Predictive Association Rules):
o Combines associative classification with predictive power.

Page 58 of 71
➢ MODULE 6
Q1) What is Text Mining?
Answer:
• Text Mining is the process of examining large collections of unstructured
textual data to generate new information, typically using specialized
software tools.
o Purpose: Extract useful information and knowledge hidden in the
text content.
o Techniques Involved:
▪ Categorization: The process of grouping items or texts into
categories based on shared characteristics.
▪ Entity Extraction: Identifying key entities such as names,
places, dates, or other relevant information from the text.
▪ Sentiment Analysis: Determining the sentiment or emotional
tone of the text (positive, negative, neutral).
o Text Mining techniques enable machines to understand, categorize,
and extract actionable insights from text data, making it invaluable
for various industries.
Q2) Why is Text Mining important?
Answer:
• Unstructured Data:
o Approximately 90% of the world’s data is stored in unstructured
formats.
o Common sources of unstructured data include:
▪ Web pages
▪ Emails
▪ Technical documents
▪ Corporate documents
▪ Books
▪ Digital libraries

Page 59 of 71
 ▪ Customer complaint letters
o This data is growing rapidly and is crucial for businesses,
governments, and other organizations to manage effectively.
o Text Mining is important as it helps in transforming this unstructured
data into valuable, structured insights, facilitating better decision-
making, enhancing operational efficiencies, and driving innovations.
Q3) What are some applications of Text Mining?
Answer:
• Spam Filtering: Text Mining techniques are used to detect spam emails by
analyzing the content and filtering out unwanted messages.
• Social Media Data Analysis: Analyzing social media posts to extract
trends, sentiments, and user opinions that can help businesses or
individuals in decision-making.
• Risk Management: Identifying and mitigating potential risks by mining
textual data from various sources such as news, blogs, and financial
reports.
• Knowledge Management: Efficiently categorizing, storing, and retrieving
business knowledge and documents within an organization.
• Cybercrime Prevention: Detecting fraudulent activities, scams, or other
cybercrimes by analyzing texts, emails, or social media posts.
• Customer Care Service: Analyzing customer feedback, complaints, and
queries to enhance customer service and address issues promptly.
• Fraud Detection: Analyzing textual patterns in documents, emails, or
reports to identify fraudulent behavior or activities.
• Contextual Advertising: Analyzing text to identify relevant content,
helping advertisers target the right audience with ads.
• Business Intelligence: Extracting key business insights from textual data
sources such as reports, news articles, or customer feedback.
• Content Enrichment: Adding valuable metadata or tags to textual content
to make it more discoverable and usable in various contexts.
Q4) What are the characteristics of textual data?
Answer:

Page 60 of 71
 • Unstructured Text:
o This includes written resources, chatroom conversations, or natural
speech that do not follow a predefined format.
o Text Mining processes unstructured data to extract meaningful
insights.
• High Dimensionality:
o Textual data can have tens of thousands of words, making it a high-
dimensional dataset.
o However, the text is often sparse, meaning only a small portion of
the vocabulary is used in any given document.
o The challenge is managing and processing this vast space of
vocabulary.
• Complex and Subtle Relationships Between Concepts:
o Textual data can have sentence ambiguity or word ambiguity:
▪ Example 1: "AOL merges with Time-Warner" vs. "Time-
Warner is bought by AOL" — these sentences express the
same event but have different structures.
▪ Example 2: Automobile = car = vehicle = Toyota — a single
concept can be expressed with different words, making it hard
for machines to associate them.
▪ Word Sense Disambiguation: For instance, "Apple" can
refer to the company or the fruit, and Text Mining must
identify the correct context to interpret it accurately.
• Noisy Data:
o Textual data often contains spelling mistakes, incorrect grammar,
or irrelevant words that need to be cleaned during the preprocessing
phase to ensure accuracy.
Q5) What is the Text Mining Process?
Answer:
• The Text Mining Process involves several steps to preprocess and
transform unstructured text into structured data that can be analyzed. These
steps include:

Page 61 of 71
 o Text Preprocessing: Cleaning and preparing the text data for
analysis by removing unwanted elements and standardizing the
format.
o Feature Generation: Converting the cleaned text into numerical
features for machine learning models.
o Text Classification: Assigning labels or categories to documents
based on their content.
Each step is crucial in ensuring that the text data is in the right format for further
mining and analysis.

Q6) What are the common techniques for Text Preprocessing?

Answer:
• Text Cleanup:
o Remove unnecessary information such as ads from web pages or
special characters (e.g., @, #, !).
o Normalize text by converting slang, abbreviations, and contractions
into their full forms (e.g., “can't” → “cannot”).
o Convert text to lowercase, remove punctuation, numbers, and
unnecessary whitespaces.
• Stopword Removal:
o Remove words that offer little semantic value, such as "a", "and",
"the", etc.

Page 62 of 71
 o Custom stopword lists may be created depending on the application
or domain.

• Stemming:
o Reduces a word to its root form by stripping away prefixes or
suffixes (e.g., "running" → "run").
o Stemming is often faster and simpler but may result in non-standard
forms of words.

• Lemmatization:
o Converts words to their base or dictionary form by considering their
context (e.g., "better" → "good").
o More accurate than stemming as it takes into account the
grammatical context.

Page 63 of 71
 • Tokenization:
o The process of breaking down text into smaller components such as
words, phrases, or symbols.
o Part-of-speech tagging is used to categorize tokens into
grammatical categories like nouns, verbs, etc.
• Parsing:
o Involves analyzing the syntactic structure of sentences to identify
relationships between words and their roles in the sentence.

Q7) How is text data represented for analysis?

Answer:
• Feature Generation:
o Text is transformed into numerical features for use in machine
learning models. Common methods include:
▪ Bag of Words (BoW): Represents text as a frequency
distribution of words, where each word is counted and used
as a feature.
▪ Vector Space Model: Represents documents as vectors,
where each component corresponds to a unique term in the
corpus, weighted by frequency or importance (using methods
like TF-IDF).
• Feature Selection:
o Reduces the dimensionality of the text data by selecting only the
most relevant features for the analysis.
o Methods such as TF-IDF help in determining the importance of
words by considering their frequency across documents, and down-
weighting common words.

Page 64 of 71
Q8) What is Web Mining?
Answer:
• Web Mining is the process of mining data related to the World Wide Web.
o Rapid Growth: The web is growing and changing at a fast pace,
making it a dynamic source of information.
o Diversity of User Communities: The web serves a broad diversity
of users, which adds complexity to the data being mined.
o Largest Database: The web contains a massive amount of data, but
much of it is unstructured or irrelevant.
o Challenges:
▪ A lot of the information on the web is useless for most users.
▪ It’s difficult to find relevant and high-quality web pages on a
specific topic.
▪ Quality vs. Quantity: The challenge lies in identifying high-
quality, useful information in the midst of an overwhelming
amount of irrelevant data.
Q9) What are the different types of Web Data?
Answer:
Web Mining works with several types of data that are available on the web,
including:
• Content of Actual Web Pages: The textual, visual, or multimedia content
present on web pages that can be analyzed for various purposes.
• Intrapage Linkage Structure: The links that exist within a single web
page connecting different parts or sections.
• Interpage Linkage Structure: The links between different web pages,
which form the hypertext structure of the web.
• Usage Data: This includes information about the pages accessed by users,
such as clickstream data.
• User Profile: This involves the demographic and personal data of users,
such as their registration details, which can be used for personalization and
targeted advertising.

Page 65 of 71
Q10) What are the different types of Web Mining?
Answer:
Web Mining is categorized into three types:
• Web Content Mining:
o Focuses on extracting useful information from the content of web
pages, including text, images, videos, and other multimedia
elements.
o Techniques used are similar to Text Mining, where the goal is to
mine data from the textual content on the web.
• Web Structure Mining:
o Involves analyzing the structure of the web and the interconnections
between web pages.
o It uses graph theory to study the organization of web pages and
their connections (links).
• Web Usage Mining:
o Focuses on analyzing web logs to understand user behavior and web
usage patterns.
o It helps in discovering useful insights from the browsing behavior of
users, often for the purpose of website optimization, personalization,
and targeted content delivery.
Q11) What is Web Content Mining?
Answer:
• Web Content Mining is a subfield of web mining that deals with
extracting useful data from the content found on web pages.
o Traditional Web Searching: Traditional search engines primarily
focus on searching and retrieving web pages based on keywords.
o Search Engine Process:
▪ Crawlers are used to search the web and gather information from
web pages.
▪ The gathered data is stored using indexing techniques to facilitate
efficient and quick retrieval of information.

Page 66 of 71
 ▪ When a user submits a query, the search engine processes the query
and retrieves relevant web pages, providing fast and accurate
results.
o Applications:
▪ Search engines use web content mining techniques to rank web
pages based on the relevance of their content to the search query.
▪ Crawlers: These tools visit websites, index content, and store it
for later retrieval when users perform searches.
Q12) What is Web Structure Mining?
Answer:
• Web Structure Mining is the process of using the structure of the web (the
connections and links between pages) to extract useful information.
o Graph Theory: It applies graph theory to analyze the nodes (web
pages) and edges (links) in the hypertext structure of the web.
o Applications:
▪ Web structure mining is used to classify web pages or to create
similarity measures between documents.
▪ It helps in understanding how web pages are organized and
how they relate to one another, which is useful for tasks like
ranking and clustering.
Q13) What is Web Usage Mining?
Answer:
• Web Usage Mining involves the mining of web log data to uncover
patterns in user behavior while they navigate websites.
o Web Logs: These logs capture clickstream data, which provides
insights into the pages users visit and how they interact with a
website.
o Client vs. Server Perspective:
▪ Server Perspective: Mining data from web server logs
reveals information about the sites where the server resides
and user navigation patterns.

Page 67 of 71
 ▪ Client Perspective: Data is collected from the user's
interactions, which may include personal information such as
demographics and browsing behavior.
o Applications:
▪ Personalization: Web Usage Mining helps personalize the
user experience by recommending content based on the user’s
past browsing behavior.
▪ Web Design Improvement: Analyzing usage patterns allows
web designers to optimize site structure and navigation for
better user experience.
Q14) What are the techniques used in Web Usage Mining?
Answer:
Several data mining techniques are applied in Web Usage Mining:
• Association Rule Mining:
o This technique finds relationships between web pages that
frequently appear together in user sessions.
o Example: Identifying pages that are often viewed together in a
sequence and recommending them to other users for cross-selling or
navigation improvements.
• Sequential Patterns:
o Involves identifying frequent navigation sequences by users.
o It helps in understanding the typical paths users take through a
website, which can aid in improving site layout or content
recommendations.
• Clustering:
o Users or pages are grouped into clusters based on similarities in
behavior or attributes.
o This is useful for partitioning markets in e-commerce or identifying
similar content on a website.
• Classification:

Page 68 of 71
 o This technique involves grouping users based on demographics or
behavior patterns, helping in the development of targeted marketing
strategies.
o For example, classifying clients based on their interaction patterns
or demographic information.
Q15) What are the applications of Web Usage Mining?
Answer:
Web Usage Mining has several key applications:
• Personalization:
o It helps in delivering a personalized web experience by adapting
content and recommendations based on individual user behavior.
o It enhances user engagement by providing more relevant content or
ads.
• Website Performance Improvement:
o By analyzing frequent access behavior, web designers can improve
the website layout, structure, and content delivery.
o Frequent access patterns can be used to enhance server caching and
speed up access to popular pages.
• Linkage Structure Modifications:
o Understanding common access behaviors allows website
administrators to adjust the internal linking structure, ensuring users
can easily navigate the most important content.
• Business Intelligence:
o Web Usage Mining is used to gather intelligence about customer
behavior, which can be used to improve business strategies, such as
enhancing sales tactics or optimizing advertisements.
Q16) What are the different types of Crawlers used in Web Mining?
Answer:
Crawlers, also known as spiders or spiderbots, are programs used to collect data
from the web. There are different types:
• Periodic Crawlers:

Page 69 of 71
 o These are activated at regular intervals and update the web index by
replacing the previous index with new content.
• Incremental Crawlers:
o These update the index incrementally instead of replacing the entire
index, making them more efficient for frequent updates.
• Focused Crawlers:
o These visit only pages related to specific topics of interest, making
them more efficient for topic-specific data mining and reducing
unnecessary resource consumption.

Q17) How do Focused Crawlers differ from Regular Crawlers?

Answer:
• Focused Crawler:
o It only visits web pages related to a specific topic or subject of
interest.
o It avoids irrelevant pages and focuses on gathering data from the
most relevant sources.
o More scalable and efficient, as it does not waste resources on
irrelevant pages.
• Regular Crawler:
o It visits every page on the web, regardless of relevance.
o It can visit more pages but is less efficient since it includes many
irrelevant pages.

Page 70 of 71
Page 71 of 71