Essay Questions
1. a) Define Data Mining and provide a detailed classification of the types of data that can be mined (Database, Data
Warehouse, Transactional, etc.).
OR
b) Explain the Data Cleaning and Data Integration processes in detail, providing specific techniques for handling missing
values and redundant data.
2. a) Draw and explain the 3-tier Data Warehouse Architecture. Discuss the role of each tier in the KDD process.
OR
b) Compare and contrast OLAP vs. OLTP systems. Provide a detailed note on Data Warehouse using Fact
Constellation Schema.
3. a) Describe the Apriori Algorithm. Illustrate the process of finding frequent item sets with a sample dataset and
candidate generation steps
. OR
b) What is a Pattern Growth approach? Explain the FP-Growth algorithm and how it avoids candidate generation.
4. a) Explain the construction of a Decision Tree using the Information Gain or Gini Index measures.
OR
b) Discuss Bayesian Classification. Explain how Bayes' Theorem is used to predict class membership probabilities.
5. a) Compare Partitioning methods: Explain the working of the K-Means algorithm and contrast it with K-Medoids.
OR
b) Explain the DBSCAN clustering algorithm. Define core points, border points, and noise in the context of density-
based clustering.
Short Questions
6. Explain the concept of Data Discretization and its importance in preprocessing.
7. Write a note on Concept Hierarchies with an example from a "Location" dimension.
8. Describe the Snowflake Schema and how it differs from the Star Schema.
9. Explain the "Slice" and "Dice" operations in a Multidimensional Data Cube.
10. Define the measures of "Interestingness" in association rule mining.
11. What is the Holdout Method in classifier evaluation?
12. Briefly explain the difference between Agglomerative and Divisive hierarchical clustering.
, DATA WAREHOUSING AND DATA MINING: MODEL PAPER 1SECTION A: ESSAY QUESTIONS (5 X
7M = 35 MARKS)
1. a) Define Data Mining and provide a detailed classification of the types of data that can be
mined.
The Concept of Knowledge Discovery Data Mining is the process of extracting hidden,
valid, and actionable information from large databases. It is a crucial step in the
Knowledge Discovery from Data (KDD) process. In an era where organizations are "Data
Rich" but "Knowledge Poor," mining acts as the bridge that turns raw bits into strategic
business intelligence.
Relational Databases & Data Ware houses
A Relational Database is a collection of tables, each with a unique key. Mining here focuses on
trends such as customer behavior. A Data Warehouse, however, integrates data from multiple
sources into a single schema. It allows for "Multidimensional Mining," where a manager can
analyze sales data across different dimensions like Time, Region, and Product Category
simultaneously.
Transactional & Stream Data
Transactional data consists of records representing events (e.g., a retail receipt). This is the
foundation for Market Basket Analysis. Data Streams refer to continuous flows of data, such
as website clickstreams, sensor data from IoT devices, or stock market tickers. Mining
streams requires real-time algorithms that can handle data that is too fast to be stored
permanently.
Advanced & Multimedia
Data Modern mining extends beyond text and numbers. Spatial Data includes maps and
satellite imagery used for geographic trends. Time-Series Data tracks changes over time, like
medical heart rate monitors. Multimedia Data involves mining images, video, and audio for
pattern recognition. Finally, the World Wide Web presents a massive repository of semi-
structured data where we mine hyperlinks (Link Analysis) and page content.
Strategic Conclusion :
The ability to mine diverse data types allows a business to build a 360-degree view of its
operations. Whether it is predicting a machine failure via sensor streams or identifying a
buying trend through social media mining, the goal remains the same: transforming raw data
into a competitive advantage.
or
1. b) Explain the Data Preprocessing steps: Data Cleaning and Data Integration with illustrative
examples.
The Necessity of Preprocessing Raw data is often "dirty"—it contains missing values,
noise, and inconsistencies. If the input data is of poor quality, the resulting patterns will be
,misleading. This is known as "Garbage In, Garbage Out." Preprocessing is the most time-
consuming yet critical phase of any data mining project, ensuring that the data is accurate,
complete, and consistent.
Data Cleaning: Handling Missing Values & Noise Data Cleaning attempts to "fill the gaps"
and remove errors. Handling Missing Values: If a customer’s age is missing, we can fill it with
a global constant, the attribute mean, or the most probable value using Bayesian inference.
Smoothing Noisy Data: Noise is a random error or variance in a measured variable. We use
"Binning" (grouping data into small segments and smoothing by mean), "Regression" (fitting
data to a function), or "Outlier Analysis" (using clustering to find and remove data points that
don't belong).
Data Integration: Resolving Conflicts Integration involves merging data from multiple
sources into a single store. Entity Identification: Ensuring that "Emp_ ID" in the HR database
and "Staff_ No" in the Payroll database refer to the same individual. Redundancy Analysis:
Some attributes may be redundant (e.g., "Age" and "Date of Birth"). We use the $Chi-square$
test for categorical data to detect correlations. If two attributes are highly correlated, one can
be removed to save space and improve algorithm speed.
Data Transformation & Discretization Transformation converts data into forms suitable for
mining. This includes "Normalization," where we scale all numerical values to a range like [0.0
to 1.0] so that large numbers don't dominate the results. Discretization involves replacing raw
values with labels, such as changing specific ages (21, 25, 29) into categories like "Young
Adult," which simplifies the mining process.
Final Insight Preprocessing is not just a cleaning task; it is a data enrichment process. By
resolving conflicts and smoothing noise, we create a high-quality "Data Cube" or "Flat File"
that allows mining algorithms to run faster and produce results that a business can actually
trust for decision-making.
2. a) Draw and explain the 3-tier Data Warehouse Architecture. (7 Marks)
The Architectural Concept A Data Warehouse is a multi-layered environment designed to
transform raw data into actionable intelligence. The Three-Tier Architecture is the standard
framework that ensures scalability and performance by separating the processes of data
storage, multidimensional processing, and user interaction.
, \
The Bottom Tier: Warehouse Database Server This is the storage layer where the physical
data resides. It is typically a Relational Database Management System (RDBMS).
• Data Sources: It gathers data from operational databases (OLTP) and external sources.
• ETL Tools: It uses Back-End tools for Extraction, Transformation, and Loading (ETL) to
clean and prepare the data before storage.
1. a) Define Data Mining and provide a detailed classification of the types of data that can be mined (Database, Data
Warehouse, Transactional, etc.).
OR
b) Explain the Data Cleaning and Data Integration processes in detail, providing specific techniques for handling missing
values and redundant data.
2. a) Draw and explain the 3-tier Data Warehouse Architecture. Discuss the role of each tier in the KDD process.
OR
b) Compare and contrast OLAP vs. OLTP systems. Provide a detailed note on Data Warehouse using Fact
Constellation Schema.
3. a) Describe the Apriori Algorithm. Illustrate the process of finding frequent item sets with a sample dataset and
candidate generation steps
. OR
b) What is a Pattern Growth approach? Explain the FP-Growth algorithm and how it avoids candidate generation.
4. a) Explain the construction of a Decision Tree using the Information Gain or Gini Index measures.
OR
b) Discuss Bayesian Classification. Explain how Bayes' Theorem is used to predict class membership probabilities.
5. a) Compare Partitioning methods: Explain the working of the K-Means algorithm and contrast it with K-Medoids.
OR
b) Explain the DBSCAN clustering algorithm. Define core points, border points, and noise in the context of density-
based clustering.
Short Questions
6. Explain the concept of Data Discretization and its importance in preprocessing.
7. Write a note on Concept Hierarchies with an example from a "Location" dimension.
8. Describe the Snowflake Schema and how it differs from the Star Schema.
9. Explain the "Slice" and "Dice" operations in a Multidimensional Data Cube.
10. Define the measures of "Interestingness" in association rule mining.
11. What is the Holdout Method in classifier evaluation?
12. Briefly explain the difference between Agglomerative and Divisive hierarchical clustering.
, DATA WAREHOUSING AND DATA MINING: MODEL PAPER 1SECTION A: ESSAY QUESTIONS (5 X
7M = 35 MARKS)
1. a) Define Data Mining and provide a detailed classification of the types of data that can be
mined.
The Concept of Knowledge Discovery Data Mining is the process of extracting hidden,
valid, and actionable information from large databases. It is a crucial step in the
Knowledge Discovery from Data (KDD) process. In an era where organizations are "Data
Rich" but "Knowledge Poor," mining acts as the bridge that turns raw bits into strategic
business intelligence.
Relational Databases & Data Ware houses
A Relational Database is a collection of tables, each with a unique key. Mining here focuses on
trends such as customer behavior. A Data Warehouse, however, integrates data from multiple
sources into a single schema. It allows for "Multidimensional Mining," where a manager can
analyze sales data across different dimensions like Time, Region, and Product Category
simultaneously.
Transactional & Stream Data
Transactional data consists of records representing events (e.g., a retail receipt). This is the
foundation for Market Basket Analysis. Data Streams refer to continuous flows of data, such
as website clickstreams, sensor data from IoT devices, or stock market tickers. Mining
streams requires real-time algorithms that can handle data that is too fast to be stored
permanently.
Advanced & Multimedia
Data Modern mining extends beyond text and numbers. Spatial Data includes maps and
satellite imagery used for geographic trends. Time-Series Data tracks changes over time, like
medical heart rate monitors. Multimedia Data involves mining images, video, and audio for
pattern recognition. Finally, the World Wide Web presents a massive repository of semi-
structured data where we mine hyperlinks (Link Analysis) and page content.
Strategic Conclusion :
The ability to mine diverse data types allows a business to build a 360-degree view of its
operations. Whether it is predicting a machine failure via sensor streams or identifying a
buying trend through social media mining, the goal remains the same: transforming raw data
into a competitive advantage.
or
1. b) Explain the Data Preprocessing steps: Data Cleaning and Data Integration with illustrative
examples.
The Necessity of Preprocessing Raw data is often "dirty"—it contains missing values,
noise, and inconsistencies. If the input data is of poor quality, the resulting patterns will be
,misleading. This is known as "Garbage In, Garbage Out." Preprocessing is the most time-
consuming yet critical phase of any data mining project, ensuring that the data is accurate,
complete, and consistent.
Data Cleaning: Handling Missing Values & Noise Data Cleaning attempts to "fill the gaps"
and remove errors. Handling Missing Values: If a customer’s age is missing, we can fill it with
a global constant, the attribute mean, or the most probable value using Bayesian inference.
Smoothing Noisy Data: Noise is a random error or variance in a measured variable. We use
"Binning" (grouping data into small segments and smoothing by mean), "Regression" (fitting
data to a function), or "Outlier Analysis" (using clustering to find and remove data points that
don't belong).
Data Integration: Resolving Conflicts Integration involves merging data from multiple
sources into a single store. Entity Identification: Ensuring that "Emp_ ID" in the HR database
and "Staff_ No" in the Payroll database refer to the same individual. Redundancy Analysis:
Some attributes may be redundant (e.g., "Age" and "Date of Birth"). We use the $Chi-square$
test for categorical data to detect correlations. If two attributes are highly correlated, one can
be removed to save space and improve algorithm speed.
Data Transformation & Discretization Transformation converts data into forms suitable for
mining. This includes "Normalization," where we scale all numerical values to a range like [0.0
to 1.0] so that large numbers don't dominate the results. Discretization involves replacing raw
values with labels, such as changing specific ages (21, 25, 29) into categories like "Young
Adult," which simplifies the mining process.
Final Insight Preprocessing is not just a cleaning task; it is a data enrichment process. By
resolving conflicts and smoothing noise, we create a high-quality "Data Cube" or "Flat File"
that allows mining algorithms to run faster and produce results that a business can actually
trust for decision-making.
2. a) Draw and explain the 3-tier Data Warehouse Architecture. (7 Marks)
The Architectural Concept A Data Warehouse is a multi-layered environment designed to
transform raw data into actionable intelligence. The Three-Tier Architecture is the standard
framework that ensures scalability and performance by separating the processes of data
storage, multidimensional processing, and user interaction.
, \
The Bottom Tier: Warehouse Database Server This is the storage layer where the physical
data resides. It is typically a Relational Database Management System (RDBMS).
• Data Sources: It gathers data from operational databases (OLTP) and external sources.
• ETL Tools: It uses Back-End tools for Extraction, Transformation, and Loading (ETL) to
clean and prepare the data before storage.
