Advanced Data Mining – Tricky MCQs (Lecture 1)

1️⃣ Data mining mainly focuses on:

A) Storing huge volumes of data efficiently

B) Extracting useful and unexpected patterns from large datasets
C) Compressing data for faster storage
D) Deleting irrelevant data automatically

✅ Answer: B

2️⃣ The most appropriate description of data mining is:

A) Converting raw data into formatted tables

B) Summarizing structured databases only
C) Discovering useful patterns and relationships in large observational datasets
D) Visualizing data using graphs and dashboards

✅ Answer: C

3️⃣ In the data mining workflow, which stage normally comes before modeling?

A) Deployment
B) Data preparation
C) Evaluation
D) Feature extraction

✅ Answer: B

4️⃣ Which of the following best represents the correct order in the Data Mining Life
Cycle?

A) Modeling → Data Preparation → Business Understanding → Evaluation

B) Business Understanding → Data Understanding → Data Preparation → Modeling →
Evaluation → Deployment
C) Data Preparation → Business Understanding → Modeling → Evaluation →
Deployment
D) Data Understanding → Modeling → Business Understanding → Deployment

✅ Answer: B

5️⃣ The main reason data mining became important in modern computing is:

A) Increase in processor speed

B) Increase in raw data generation and cheap storage
C) Decrease in database sizes
D) Reduction in algorithm complexity

✅ Answer: B

6️⃣ Which of the following is NOT a major motivation for data mining?

A) Predictive analytics
B) Knowledge discovery
C) Eliminating all human decision-making
D) Decision support

✅ Answer: C

7️⃣ Discovering relationships between variables that were previously unknown is

referred to as:

A) Data cleaning
B) Knowledge discovery
C) Data aggregation
D) Dimensional reduction

✅ Answer: B
8️⃣ A predictive model developed using historical data is primarily used for:

A) Data compression
B) Forecasting future outcomes
C) Improving database indexing
D) Reducing dataset size

✅ Answer: B

9️⃣ Which of the following best represents pattern recognition in data mining?

A) Detecting errors in databases

B) Grouping similar data objects based on characteristics
C) Storing similar data in same memory location
D) Converting text into numbers

✅ Answer: B

🔟 Which statement best explains competitive advantage through data mining?

A) Companies store more data than competitors

B) Companies analyze data to better understand markets and customers
C) Companies increase computing hardware
D) Companies remove outdated datasets

✅ Answer: B

Data Types MCQs

11️⃣ In data mining terminology, an object refers to:

A) A property of a dataset
B) A row containing a set of attributes
C) A statistical model
D) A feature extraction algorithm

✅ Answer: B

12️⃣ An attribute is best defined as:

A) A data mining algorithm

B) A characteristic or property describing an object
C) A group of similar objects
D) A stored data file

✅ Answer: B

13️⃣ Which of the following is a nominal attribute?

A) Temperature
B) Height
C) Marital Status
D) Age

✅ Answer: C

14️⃣ Which property distinguishes ordinal data from nominal data?

A) Ordinal data supports mathematical operations

B) Ordinal data has meaningful ranking
C) Ordinal data is numeric only
D) Ordinal data contains continuous values

✅ Answer: B
15️⃣ Which of the following is the best example of discrete data?

A) Temperature
B) Distance travelled
C) Number of students in a class
D) Weight of a person

✅ Answer: C

16️⃣ Continuous data differs from discrete data because it:

A) Represents categories only

B) Can take any value within a range
C) Always uses integers
D) Cannot be measured

✅ Answer: B

Data Types & Representations

17️⃣ In document representation using bag-of-words, the order of words:

A) Is extremely important
B) Is ignored
C) Is encoded numerically
D) Determines cluster membership

✅ Answer: B

18️⃣ Transaction data can be represented as:

A) A binary vector of items

B) A sequence of timestamps
C) A graph structure
D) A time series

✅ Answer: A

19️⃣ Time-series data is characterized by:

A) Unordered attributes
B) Random attribute relationships
C) Ordered numeric values over time
D) Binary item representation

✅ Answer: C

20️⃣ Web pages connected through hyperlinks represent which type of data?

A) Ordered data
B) Transaction data
C) Graph data
D) Numerical vector data

✅ Answer: C

Data Mining Tasks

21️⃣ Discovering products frequently purchased together is known as:

A) Classification
B) Clustering
C) Association rule mining
D) Regression

✅ Answer: C
22️⃣ The rule {Milk, Diaper} → {Beer} represents:

A) A classification model
B) An association rule
C) A clustering result
D) A regression equation

✅ Answer: B

23️⃣ In clustering, the goal is to:

A) Maximize similarity between clusters

B) Minimize similarity within clusters
C) Maximize similarity within clusters and minimize similarity between clusters
D) Assign predefined class labels

✅ Answer: C

24️⃣ Which distance metric is commonly used for clustering continuous attributes?

A) Manhattan distance
B) Euclidean distance
C) Hamming distance
D) Binary distance

✅ Answer: B

25️⃣ In classification problems, the dataset used to train the model is called:

A) Test set
B) Training set
C) Validation set
D) Evaluation set
✅ Answer: B

26️⃣ Which dataset is used to measure the accuracy of a classification model?

A) Training dataset
B) Test dataset
C) Transaction dataset
D) Feature dataset

✅ Answer: B

Applications of Data Mining

27️⃣ Predicting whether a user will click an online advertisement is an example of:

A) Clustering
B) Classification
C) Association rule mining
D) Graph mining

✅ Answer: B

28️⃣ Credit card fraud detection mainly uses:

A) Clustering
B) Classification models
C) Frequent itemsets
D) Time series clustering

✅ Answer: B

29️⃣ Determining the importance of web pages based on links is related to:
A) Classification
B) Ranking using link analysis
C) Clustering
D) Data summarization

✅ Answer: B

30️⃣ Predicting new connections in a social network relies on the concept of:

A) Data compression
B) Triadic closure
C) Attribute independence
D) Dimensional reduction

✅ Answer: B

Conceptual Understanding

31️⃣ Data mining is considered a convergence of which disciplines?

A) Networking, cryptography, operating systems

B) Machine learning, statistics, and databases
C) Hardware design, AI, and operating systems
D) Computer graphics and networking

✅ Answer: B

32️⃣ Traditional statistical methods often struggle with modern data mining problems
because of:

A) Low data availability

B) High dimensional and heterogeneous data
C) Lack of programming languages
D) Small dataset sizes
✅ Answer: B

33️⃣ Exploratory data analysis mainly aims to:

A) Predict future data values

B) Understand patterns and structure in data
C) Compress datasets
D) Eliminate irrelevant attributes

✅ Answer: B

34️⃣ A model that explains relationships within data rather than predicting outcomes is
primarily used for:

A) Prediction
B) Summarization
C) Explanation
D) Compression

✅ Answer: C

35️⃣ According to the lecture, the role of a data scientist includes:

A) Only database management

B) Only statistical modeling
C) Combining statistics, programming, and data analysis
D) Designing hardware systems

✅ Answer: C

Behavioral Data
1️⃣ Behavioral data collected from mobile devices may include:
A) Only stored contact numbers
B) GPS location, communication records, and application interactions
C) Only device hardware information
D) Processor performance metrics

✅ Answer: B

2️⃣ Which organization is most likely to collect behavioral data related to user browsing
and search queries?
A) Manufacturing company
B) Search engine provider
C) Weather monitoring center
D) Banking ATM network

✅ Answer: B

3️⃣ Behavioral data primarily helps organizations to:

A) Increase hardware storage capacity
B) Understand and predict user behavior
C) Improve network bandwidth
D) Encrypt data transactions

✅ Answer: B

Recommendation Systems
4️⃣ A recommendation system suggesting “Users who bought this also bought…” is
based on:
A) Association rules
B) Regression models
C) Data normalization
D) Time series forecasting

✅ Answer: A
5️⃣ Recommendation systems mainly rely on:
A) Network protocols
B) Similarity between users or items
C) Hardware acceleration
D) Database indexing

✅ Answer: B

6️⃣ Which company widely uses recommendation systems to suggest products?

A) Amazon
B) NASA
C) Intel
D) Cisco

✅ Answer: A

Retail Market Basket Analysis

7️⃣ Market basket analysis is primarily used to:
A) Monitor customer loyalty cards
B) Identify products frequently purchased together
C) Optimize warehouse temperature
D) Track shipment logistics

✅ Answer: B

8️⃣ The classic market basket rule {Diaper, Milk → Beer} demonstrates:
A) Regression analysis
B) Sequential mining
C) Association rule mining
D) Clustering

✅ Answer: C
9️⃣ Shelf management in supermarkets benefits from:
A) Data compression techniques
B) Market basket analysis
C) Image recognition algorithms
D) Network routing

✅ Answer: B

Bioinformatics Clustering

🔟 In bioinformatics, clustering is often used to group:

A) Protein manufacturing machines
B) Genes with similar expression patterns
C) Database transactions
D) Network nodes

✅ Answer: B

11️⃣ Gene expression data typically contains:

A) Small number of attributes and huge samples
B) Thousands of genes measured across conditions
C) Only binary data values
D) Only textual information

✅ Answer: B

12️⃣ Clustering genes helps scientists to:

A) Increase storage capacity
B) Identify similar biological functions
C) Replace statistical analysis
D) Compress genomic sequences

✅ Answer: B
Stock Market Analytics
13️⃣ In financial data mining, stock prices recorded over time represent:
A) Transaction data
B) Graph data
C) Time series data
D) Binary data

✅ Answer: C

14️⃣ Clustering stocks based on price movement helps investors to:

A) Identify groups of stocks behaving similarly
B) Predict server downtime
C) Store financial records efficiently
D) Compress stock market databases

✅ Answer: A

15️⃣ Predicting future stock values is an example of:

A) Predictive analytics
B) Data compression
C) Transaction processing
D) Clustering

✅ Answer: A

Bow-Tie Web Structure

16️⃣ The bow-tie model is used to describe the structure of:
A) Social networks
B) Web graph structure
C) Financial networks
D) Wireless sensor networks

✅ Answer: B
17️⃣ In the bow-tie structure of the web, the SCC (Strongly Connected Component)
represents:
A) Isolated pages
B) Pages mutually reachable via hyperlinks
C) Pages without outgoing links
D) Pages not indexed by search engines

✅ Answer: B

18️⃣ Pages that can reach the SCC but cannot be reached from it belong to:
A) IN component
B) OUT component
C) Tendrils
D) Tubes

✅ Answer: A

Query Logs
19️⃣ Query logs maintained by search engines typically record:
A) Server hardware usage
B) User queries and clicked links
C) CPU temperature values
D) Email server messages

✅ Answer: B

20️⃣ Analyzing query logs helps search engines to:

A) Improve query suggestions and search ranking
B) Increase internet speed
C) Reduce server electricity consumption
D) Encrypt web pages
✅ Answer: A

Document Clustering
21️⃣ Document clustering groups documents based on:
A) File storage location
B) Similarity of terms or topics
C) Author's writing style only
D) Document file size

✅ Answer: B

22️⃣ In document mining, a document is usually represented as:

A) Binary graph
B) Term vector
C) Time series
D) Network node

✅ Answer: B

23️⃣ Document clustering improves:

A) Network security
B) Information retrieval systems
C) Database backup speed
D) Hardware utilization

✅ Answer: B

Web Graph Distribution

24️⃣ The distribution of incoming links to web pages follows:
A) Uniform distribution
B) Random distribution
C) Power-law distribution
D) Linear distribution

✅ Answer: C

25️⃣ In web graph analysis, pages with many incoming links are considered:
A) Less reliable
B) Highly authoritative
C) Redundant
D) Temporary

✅ Answer: B

Social Network & Link Prediction

26️⃣ Predicting future friendships in a social network is called:
A) Link prediction
B) Classification
C) Clustering
D) Regression

✅ Answer: A

27️⃣ The concept that two people with a common friend are likely to connect is called:
A) Graph partitioning
B) Triadic closure
C) Binary classification
D) Sequential mining

✅ Answer: B

Retail Analytics
28️⃣ Retail companies analyze transaction data mainly to:
A) Improve store lighting
B) Understand customer purchasing behavior
C) Reduce employee workload
D) Increase product packaging

✅ Answer: B

29️⃣ Product placement strategies in supermarkets are influenced by:

A) Clustering algorithms
B) Association rule mining
C) Network routing algorithms
D) Sorting algorithms

✅ Answer: B

30️⃣ A recommendation like “Customers who watched this movie also watched…” is
based on:
A) Association rule mining and similarity analysis
B) Time series forecasting
C) Database indexing
D) Data compression

✅ Answer: A

Data Mining MCQs (Association Rule Mining)

1.
Which data mining technique focuses on discovering relationships among variables in
large transactional datasets?
A. Regression
B. Association Rule Mining
C. Classification
D. Dimensionality Reduction
Answer: B
2.
Which of the following best describes classification in data mining?
A. Discovering relationships between items in transactions
B. Predicting continuous values from input variables
C. Assigning items into predefined categories based on training data
D. Grouping objects without predefined labels
Answer: C

3.
Which of the following algorithms is commonly used for classification problems?
A. K-Means
B. Decision Tree
C. Apriori
D. FP-Growth
Answer: B

4.
Which technique is mainly used to predict continuous numerical values?
A. Clustering
B. Association
C. Regression
D. Classification
Answer: C

5.
Which of the following algorithms is used for clustering?
A. Random Forest
B. Support Vector Machine
C. K-Means
D. Apriori
Answer: C

6.
In market basket analysis, a transaction generally represents:
A. A group of products in inventory
B. A single product category
C. A collection of items purchased together during one shopping event
D. A list of suppliers
Answer: C

7.
Which measure indicates how frequently an itemset appears in a dataset?
A. Confidence
B. Support
C. Lift
D. Accuracy
Answer: B

8.
Which metric evaluates the likelihood that item Y is purchased when item X is
purchased?
A. Lift
B. Support
C. Confidence
D. Correlation
Answer: C
9.
In association rule mining, a rule is typically expressed in which form?
A. X ↔ Y
B. X → Y
C. X = Y
D. X || Y
Answer: B

10.
What does Lift > 1 indicate about a relationship between two items?
A. They rarely appear together
B. Their co-occurrence is less than random chance
C. They appear together more often than expected by chance
D. There is no relationship between them
Answer: C

11.
Which algorithm is considered one of the earliest approaches for association rule
mining?
A. DBSCAN
B. Apriori
C. Random Forest
D. Naïve Bayes
Answer: B

12.
The Apriori algorithm mainly operates in which two phases?
A. Data normalization and prediction
B. Feature extraction and clustering
C. Frequent itemset generation and rule generation
D. Data cleaning and visualization
Answer: C

13.
What is the main limitation of the Apriori algorithm?
A. Requires labeled data
B. Inefficient with large datasets due to candidate generation
C. Cannot handle categorical data
D. Works only with numerical variables
Answer: B

14.
Which algorithm improves efficiency by avoiding candidate itemset generation?
A. Logistic Regression
B. FP-Growth
C. Apriori
D. K-Means
Answer: B

15.
The FP-Growth algorithm stores data using which structure?
A. Decision Tree
B. Hash Table
C. FP-Tree
D. Binary Tree
Answer: C

16.
In association rule mining, an itemset containing k items is known as:
A. K-transaction
B. K-rule
C. K-itemset
D. K-pattern
Answer: C

17.
Which concept represents a set of items appearing together in a transaction?
A. Cluster
B. Itemset
C. Rule
D. Dataset
Answer: B

18.
What does Support = (Transactions containing item) / (Total transactions) represent?
A. Probability of buying two items together
B. Popularity of an item in the dataset
C. Accuracy of prediction
D. Relationship strength between two rules
Answer: B

19.
Which of the following steps occurs first in association rule mining?
A. Rule evaluation
B. Data preprocessing
C. Rule generation
D. Pattern visualization
Answer: B

20.
Which researchers defined association rule mining in the 1990s?
A. Andrew Ng, Geoffrey Hinton, Yann LeCun
B. Rakesh Agrawal, Tomasz Imieliński, Arun Swami
C. Ian Goodfellow, Yoshua Bengio, Fei-Fei Li
D. Michael Jordan, Judea Pearl, Stuart Russell
Answer: B

Numerical MCQs
21.
A supermarket recorded 500 transactions. Among them 120 transactions include milk,
and 60 transactions include both milk and bread.
What is the support of {milk, bread}?
A. 0.12
B. 0.24
C. 0.50
D. 0.60
Answer: A (60 / 500)

22.
Using the same data, what is the confidence of Milk → Bread?
A. 0.50
B. 0.40
C. 0.60
D. 0.30
Answer: A (60 / 120)
23.
In 1000 transactions, 200 contain coffee and 80 contain both coffee and sugar.
What is the support of {coffee, sugar}?
A. 0.08
B. 0.20
C. 0.40
D. 0.80
Answer: A

24.
Using the same dataset, what is the confidence of Coffee → Sugar?
A. 0.20
B. 0.40
C. 0.50
D. 0.80
Answer: B (80 / 200)

25.
A store has 2000 transactions, and 300 transactions contain bread.
What is the support of bread?
A. 5%
B. 10%
C. 15%
D. 20%
Answer: C (300 / 2000)

26.
If 100 transactions contain both bread and jam and 200 contain jam, what is the
confidence of Jam → Bread?
A. 25%
B. 50%
C. 75%
D. 100%
Answer: B

27.
In 100 transactions, 40 contain bread and 30 contain bread and butter.
What is the support of {bread, butter}?
A. 0.30
B. 0.40
C. 0.75
D. 0.10
Answer: A

28.
Using the same dataset, what is the confidence of Bread → Butter?
A. 30%
B. 40%
C. 75%
D. 90%
Answer: C

29.
If Lift = 1, what does it indicate?
A. Items are negatively related
B. Items occur independently
C. Items always appear together
D. Items never appear together
Answer: B
30.
If Lift < 1, what does it suggest?
A. Strong positive association
B. No association
C. Negative relationship between items
D. Perfect correlation
Answer: C

31.
A dataset contains 400 transactions, 80 contain item A, and 20 contain both A and B.
What is the confidence of A → B?
A. 0.25
B. 0.50
C. 0.20
D. 0.80
Answer: A

32.
Which metric controls the popularity of item Y when measuring association with X?
A. Support
B. Confidence
C. Lift
D. Accuracy
Answer: C

33.
Which algorithm uses a level-wise search strategy to generate frequent itemsets?
A. FP-Growth
B. Apriori
C. DBSCAN
D. K-Means
Answer: B

34.
Which structure is built in FP-Growth to store transaction data compactly?
A. Hash Table
B. Graph Tree
C. FP-Tree
D. B-Tree
Answer: C

35.
If the minimum support threshold is increased, what happens?
A. More frequent itemsets are generated
B. Fewer itemsets qualify as frequent
C. All itemsets become frequent
D. Support value becomes irrelevant
Answer: B

1.
Data mining can best be defined as:
A. Storing large amounts of data in databases
B. Extracting useful patterns and knowledge from large datasets
C. Compressing datasets to reduce storage
D. Converting raw data into tables
Answer: B

2.
Which technique is primarily used to discover relationships among items in large
databases?
A. Regression
B. Clustering
C. Association Rule Mining
D. Dimensionality Reduction
Answer: C

3.
Classification belongs to which type of learning?
A. Unsupervised learning
B. Semi-supervised learning
C. Supervised learning
D. Reinforcement learning
Answer: C

4.
Clustering is categorized as:
A. Supervised learning
B. Unsupervised learning
C. Reinforcement learning
D. Semi-supervised learning
Answer: B

5.
Regression is mainly used for predicting:
A. Categories
B. Continuous numeric values
C. Binary classes
D. Item associations
Answer: B

6.
An itemset is defined as:
A. A collection of related algorithms
B. A set of items occurring together in a transaction
C. A group of clusters
D. A group of databases
Answer: B

7.
A k-itemset refers to:
A. k number of transactions
B. k number of attributes
C. An itemset containing k items
D. k clusters
Answer: C

8.
A frequent itemset is one that:
A. Appears exactly once in a dataset
B. Appears more than half of all transactions
C. Satisfies the minimum support threshold
D. Has maximum confidence
Answer: C
9.
Support measures:
A. The strength of correlation between attributes
B. How often an itemset appears in transactions
C. The probability of prediction error
D. The number of clusters formed
Answer: B

10.
Confidence measures:
A. Popularity of an item
B. Probability that Y appears when X appears
C. Frequency of items in a cluster
D. Probability of error
Answer: B

11.
Lift measures:
A. Popularity of an item
B. Strength of association relative to independence
C. Number of transactions in database
D. Prediction accuracy
Answer: B

12.
If Lift = 1, it means:
A. Strong association exists
B. Items are independent
C. Items never occur together
D. Items are negatively correlated
Answer: B

13.
If Lift > 1, it indicates:
A. Positive association between items
B. Negative association
C. No relationship
D. Random occurrence
Answer: A

14.
If Lift < 1, it indicates:
A. Strong relationship
B. No relationship
C. Negative association
D. Perfect correlation
Answer: C

15.
Association rule is usually written as:
A. X + Y
B. X = Y
C. X → Y
D. X / Y
Answer: C

Part 2: Conceptual & Tricky MCQs

16.
Market Basket Analysis is mainly used to:
A. Predict stock prices
B. Analyze items frequently purchased together
C. Classify customer types
D. Detect fraud transactions
Answer: B

17.
Which of the following algorithms is most associated with association rule mining?
A. Apriori
B. K-Means
C. Linear Regression
D. Decision Tree
Answer: A

18.
The Apriori algorithm is based on the principle that:
A. All subsets of a frequent itemset must also be frequent
B. All supersets of frequent itemsets are frequent
C. All items must appear equally
D. Frequent itemsets are always large
Answer: A

19.
Which major problem does Apriori face when dealing with large datasets?
A. Data duplication
B. Candidate itemset explosion
C. Memory leaks
D. Data corruption
Answer: B

20.
Which algorithm avoids candidate generation?
A. Apriori
B. FP-Growth
C. K-Means
D. SVM
Answer: B

21.
FP-Growth algorithm uses which data structure?
A. Binary Tree
B. FP-Tree
C. Decision Tree
D. Heap
Answer: B

22.
The FP-Tree primarily stores:
A. Clusters
B. Classification models
C. Transactional itemset relationships
D. Regression parameters
Answer: C

23.
Which step occurs first in association rule mining?
A. Rule evaluation
B. Data preprocessing
C. Rule generation
D. Pattern visualization
Answer: B

24.
Which step identifies item combinations appearing frequently?
A. Rule evaluation
B. Frequent itemset mining
C. Data cleaning
D. Data integration
Answer: B

25.
Association rules are evaluated using:
A. Accuracy, precision, recall
B. Support, confidence, lift
C. Mean, median, mode
D. Entropy, gain, variance
Answer: B

Part 3: Numerical / Calculation MCQs

26.
Total transactions = 100
Transactions containing milk = 40
Support (milk) = ?
A. 0.2
B. 0.4
C. 0.5
D. 0.8
Answer: B

27.
Transactions containing milk and bread = 20
Total transactions = 100
Support (milk ∪ bread) = ?
A. 0.20
B. 0.25
C. 0.40
D. 0.50
Answer: A

28.
Milk transactions = 40
Milk & bread transactions = 20
Confidence (Milk → Bread) = ?
A. 0.2
B. 0.4
C. 0.5
D. 0.8
Answer: C

29.
If Support(X,Y)=0.3 and Support(X)=0.5
Confidence (X → Y) = ?
A. 0.6
B. 0.3
C. 0.8
D. 0.2
Answer: A

30.
If confidence of rule A → B is 80%, it means:
A. 80% of transactions contain A and B
B. 80% of transactions containing A also contain B
C. 80% of transactions contain B
D. A always occurs with B
Answer: B

Part 4: Advanced / Tricky Exam MCQs

31.
Which technique groups objects based on similarity without labels?
A. Regression
B. Clustering
C. Classification
D. Association
Answer: B

32.
Which classification algorithm uses multiple decision trees?
A. Random Forest
B. Apriori
C. DBSCAN
D. K-Means
Answer: A

33.
Support is also referred to as:
A. Probability
B. Frequency
C. Correlation
D. Accuracy
Answer: B

34.
Which measure indicates rule reliability?
A. Confidence
B. Support
C. Lift
D. Entropy
Answer: A

35.
Which measure indicates rule interestingness?
A. Lift
B. Accuracy
C. Variance
D. Recall
Answer: A

36.
The Apriori algorithm uses which strategy?
A. Divide and conquer
B. Greedy search
C. Level-wise search
D. Depth-first search
Answer: C

37.
FP-Growth mainly improves performance by:
A. Reducing support threshold
B. Avoiding candidate generation
C. Increasing dataset size
D. Reducing confidence
Answer: B

38.
Which data type is most commonly used in association rule mining?
A. Transactional data
B. Image data
C. Audio data
D. Streaming data
Answer: A

39.
Which algorithm is more efficient for large datasets?
A. Apriori
B. FP-Growth
C. Decision Tree
D. Logistic Regression
Answer: B
40.
Which concept describes the set of purchased items in a single purchase event?
A. Cluster
B. Itemset
C. Transaction
D. Rule
Answer: C

41.
Which step converts frequent itemsets into rules?
A. Rule generation
B. Clustering
C. Data normalization
D. Feature extraction
Answer: A

42.
Which value determines if an itemset is frequent?
A. Minimum support threshold
B. Maximum confidence threshold
C. Maximum support threshold
D. Minimum lift threshold
Answer: A

43.
Association rule mining originated mainly from:
A. Healthcare analysis
B. Supermarket sales analysis
C. Financial forecasting
D. Weather prediction
Answer: B

44.
Which researchers introduced association rule mining?
A. Andrew Ng, Hinton, LeCun
B. Agrawal, Imieliński, Swami
C. Bishop, Mitchell, Jordan
D. Vapnik, Chervonenkis, Quinlan
Answer: B

45.
The main goal of association analysis is to:
A. Reduce dataset size
B. Discover relationships between variables
C. Remove outliers
D. Normalize data
Answer: B

46.
Which metric considers popularity of item Y when evaluating rule X → Y?
A. Support
B. Confidence
C. Lift
D. Accuracy
Answer: C

47.
Which technique partitions data into groups based on similarity?
A. Classification
B. Clustering
C. Association
D. Regression
Answer: B

48.
Which algorithm is NOT used for association rule mining?
A. Apriori
B. FP-Growth
C. Eclat
D. K-Means
Answer: D

49.
Which step involves interpreting discovered patterns?
A. Data cleaning
B. Rule evaluation
C. Frequent itemset generation
D. Rule pruning
Answer: B

50.
Which property helps Apriori reduce search space?
A. Downward closure property
B. Bayesian property
C. Markov property
D. Gradient property
Answer: A

51–60 (Very Short Definition MCQs)

51.
The dataset in association rule mining is composed of:
A. Attributes
B. Transactions
C. Clusters
D. Trees
Answer: B

52.
Support is calculated using:
A. Transactions containing item / total transactions
B. Total items / total transactions
C. Transactions containing X / transactions containing Y
D. Total transactions / items
Answer: A

53.
Confidence is calculated using:
A. Support(X,Y) / Support(X)
B. Support(X) / Support(Y)
C. Support(X,Y) / total transactions
D. Support(Y) / Support(X,Y)
Answer: A

54.
Lift is calculated using:
A. Confidence / Support
B. Support / Confidence
C. Confidence / Support(Y)
D. Support / Support(Y)
Answer: C

55.
Which algorithm is based on divide-and-conquer?
A. Apriori
B. FP-Growth
C. Decision Tree
D. Naïve Bayes
Answer: B

56.
Which algorithm generates candidate itemsets?
A. FP-Growth
B. Apriori
C. Eclat
D. DBSCAN
Answer: B

57.
Which algorithm compresses database into a tree?
A. Apriori
B. FP-Growth
C. SVM
D. Random Forest
Answer: B
58.
Association rule mining is mostly applied in:
A. Medical imaging
B. Market basket analysis
C. Robotics
D. Speech recognition
Answer: B

59.
The strength of rule X → Y increases when:
A. Support decreases
B. Confidence increases
C. Transactions decrease
D. Dataset shrinks
Answer: B

60.
Which metric indicates how common an itemset is?
A. Support
B. Confidence
C. Lift
D. Accuracy
Answer: A

Supervised and Unsupervised Machine Learning

1. Which statement best describes supervised learning?

A. Learning patterns from unlabeled data
 B. Learning from labeled input–output pairs
C. Learning through reward signals only
D. Learning without training data
Answer: B
2. In supervised learning, the correct outputs paired with input data are called:
A. Attributes
B. Labels
C. Clusters
D. Predictions
Answer: B
3. Which of the following is an example of a classification task?
A. Predicting house prices
B. Grouping customers by behavior
C. Determining whether an email is spam or not
D. Reducing dataset dimensions
Answer: C
4. Predicting house prices based on features such as area and location is an example
of:
A. Regression
B. Clustering
C. Classification
D. Reinforcement learning
Answer: A
5. The main goal of supervised learning is to:
A. Discover hidden clusters
B. Predict outputs for unseen data
C. Compress data
D. Remove irrelevant attributes
Answer: B
6. In supervised learning datasets, input variables are also known as:
A. Dependent variables
B. Independent variables
C. Target variables
D. Labels
Answer: B
7. The output variable in supervised learning is also called:
A. Predictor
B. Feature
C. Target variable
D. Attribute
Answer: C
8. Which step involves dividing the dataset into two subsets for training and
evaluation?
A. Feature extraction
B. Data normalization
C. Training and test set division
D. Model pruning
Answer: C
9. The dataset used to train a machine learning model is called the:
A. Validation set
B. Training set
C. Test set
D. Prediction set
Answer: B
10. The purpose of the test set is to:
A. Train the model
B. Reduce dataset size
C. Evaluate model performance on unseen data
D. Remove irrelevant attributes
Answer: C
11. Feature extraction is mainly used to:
A. Increase data size
B. Reduce dataset noise and select relevant information
C. Label the dataset
D. Train models faster
Answer: B
12. Which step involves choosing the best algorithm for a machine learning problem?
A. Data preprocessing
B. Model selection
C. Prediction
 D. Evaluation
Answer: B
13. During training, the objective of the model is to:
A. Maximize dataset size
B. Minimize prediction error
C. Remove irrelevant labels
D. Increase features
Answer: B
14. Which metric is commonly used to evaluate classification models?
A. Accuracy
B. Mean squared error
C. Support
D. Lift
Answer: A
15. Linear Regression is mainly used to predict:
A. Categorical outcomes
B. Continuous numeric values
C. Clusters
D. Rewards
Answer: B
16. Logistic Regression is used when the output variable is:
A. Continuous
B. Binary or categorical
C. Numerical only
D. Unlabeled
Answer: B
17. Decision trees make predictions by:
A. Using probability distributions
B. Splitting data based on feature conditions
C. Grouping clusters
D. Random sampling
Answer: B
18. Random forests improve prediction accuracy by:
A. Removing data points
B. Combining multiple decision trees
 C. Eliminating features
D. Using unsupervised learning
Answer: B
19. Support Vector Machines create:
A. Clusters
B. Decision boundaries or hyperplanes
C. Probability tables
D. Random samples
Answer: B
20. Naive Bayes algorithm is based on:
A. Markov theorem
B. Bayes’ theorem
C. Euclidean distance
D. Gradient descent
Answer: B
21. Naive Bayes assumes that input features are:
A. Dependent
B. Independent
C. Continuous only
D. Random
Answer: B
22. K-Nearest Neighbors classifies new instances based on:
A. Distance to nearest neighbors
B. Random selection
C. Maximum entropy
D. Regression analysis
Answer: A
23. Neural networks are inspired by:
A. Computer processors
B. Human brain structure
C. Mathematical equations
D. Data compression models
Answer: B
24. Gradient boosting algorithms improve performance by:
A. Building models sequentially correcting previous errors
 B. Decreasing dataset size
C. Removing clusters
D. Ignoring misclassified samples
Answer: A
25. Spam email detection is a common application of:
A. Supervised learning
B. Reinforcement learning
C. Unsupervised learning
D. Random search
Answer: A
26. Sentiment analysis determines:
A. Numerical prediction
B. Opinion or sentiment in text
C. Cluster formation
D. Data compression
Answer: B
27. Credit scoring uses machine learning to:
A. Predict loan default risk
B. Cluster customers
C. Detect anomalies only
D. Compress financial data
Answer: A
28. A key limitation of supervised learning is:
A. Lack of algorithms
B. Dependence on labeled data
C. Low prediction accuracy
D. Inability to handle large data
Answer: B
29. Overfitting occurs when:
A. Model learns general patterns
B. Model memorizes training data
C. Model ignores training data
D. Dataset becomes smaller
Answer: B
30. Unsupervised learning works with:
A. Labeled data
B. Unlabeled data
C. Binary data only
D. Numeric data only
Answer: B
31. The main goal of unsupervised learning is to:
A. Predict labels
B. Discover patterns or structures in data
C. Minimize training error
D. Train neural networks
Answer: B
32. Clustering is used to:
A. Predict numeric outputs
B. Group similar data points
C. Reduce dimensions
D. Train reinforcement models
Answer: B
33. K-Means clustering divides data into:
A. Random groups
B. A fixed number of clusters
C. Infinite clusters
D. Binary classes
Answer: B
34. Hierarchical clustering creates:
A. Flat clusters
B. A hierarchy of clusters
C. Random clusters
D. Regression models
Answer: B
35. DBSCAN identifies clusters based on:
A. Distance thresholds only
B. Density of data points
C. Feature importance
 D. Random sampling
Answer: B
36. Dimensionality reduction is used to:
A. Increase number of features
B. Reduce number of features while keeping important information
C. Label datasets
D. Train reinforcement agents
Answer: B
37. PCA stands for:
A. Principal Component Analysis
B. Pattern Cluster Algorithm
C. Probabilistic Classification Analysis
D. Prediction Component Algorithm
Answer: A
38. t-SNE is mainly used for:
A. Classification
B. Visualization of high-dimensional data
C. Regression prediction
D. Data labeling
Answer: B
39. Association rule mining is commonly used in:
A. Image classification
B. Market basket analysis
C. Medical diagnosis
D. Robotics control
Answer: B
40. Reinforcement learning focuses on:
A. Pattern discovery
B. Prediction using labeled data
C. Learning through interaction with environment
D. Clustering data
Answer: C
41. In reinforcement learning, the decision maker is called the:
A. Model
B. Agent
 C. Cluster
D. Classifier
Answer: B
42. The system the agent interacts with is called the:
A. Dataset
B. Environment
C. Model
D. Reward system
Answer: B
43. Rewards in reinforcement learning represent:
A. Training errors
B. Positive feedback for actions
C. Input features
D. Clusters
Answer: B
44. The challenge of choosing between trying new actions and using known actions is
called:
A. Bias-variance tradeoff
B. Exploration–exploitation tradeoff
C. Overfitting problem
D. Dimensionality problem
Answer: B
45. Markov Decision Processes are used to model:
A. Sequential decision-making problems
B. Data preprocessing
C. Feature extraction
D. Clustering tasks
Answer: A
46. Value functions estimate:
A. Dataset size
B. Expected future rewards
C. Model parameters
D. Cluster numbers
Answer: B
 47. Policies in reinforcement learning determine:
A. Feature selection
B. Action selection strategies
C. Cluster formation
D. Data labeling
Answer: B
48. Q-learning is a:
A. Model-based method
B. Model-free reinforcement learning algorithm
C. Supervised learning method
D. Clustering algorithm
Answer: B
49. Deep Q-Networks combine reinforcement learning with:
A. Decision trees
B. Deep neural networks
C. K-means clustering
D. PCA
Answer: B
50. Reinforcement learning is commonly applied in:
A. Game playing and robotics
B. Data compression
C. Clustering only
D. Regression tasks
Answer: A

Machine Learning Lecture – Exam MCQs

Types of Machine Learning

1️⃣ Machine learning methods are primarily categorized into:

A) Predictive, Descriptive, Prescriptive

B) Supervised, Unsupervised, Reinforcement
C) Structured, Semi-structured, Unstructured
D) Classification, Regression, Clustering
✅ Answer: B

2️⃣ Which learning paradigm requires labeled training data?

A) Reinforcement learning
B) Supervised learning
C) Unsupervised learning
D) Semi-supervised learning

✅ Answer: B

3️⃣ The main objective of supervised learning is to:

A) Discover hidden patterns without labels

B) Learn mapping between inputs and outputs
C) Reduce dimensionality of data
D) Identify anomalies in datasets

✅ Answer: B

Supervised Learning

4️⃣ Which of the following best describes supervised learning?

A) Learning from reward signals only

B) Learning patterns from unlabeled datasets
C) Learning from input-output labeled examples
D) Learning from environmental feedback

✅ Answer: C

5️⃣ In supervised learning, the dependent variable is also called:

A) Feature
B) Predictor
C) Target label
D) Attribute

✅ Answer: C

6️⃣ Which task predicts continuous numeric values?

A) Classification
B) Regression
C) Clustering
D) Association

✅ Answer: B

7️⃣ Spam detection is an example of:

A) Regression
B) Clustering
C) Classification
D) Reinforcement learning

✅ Answer: C

Supervised Learning Process

8️⃣ Which stage involves dividing data into training and testing sets?

A) Feature extraction
B) Data preprocessing
C) Model validation
D) Data splitting

✅ Answer: D
9️⃣ The training dataset is primarily used to:

A) Evaluate model performance

B) Tune hyperparameters
C) Learn patterns and relationships
D) Visualize data

✅ Answer: C

🔟 The test dataset helps to evaluate:

A) Training speed
B) Model generalization ability
C) Feature extraction quality
D) Data labeling accuracy

✅ Answer: B

Feature Extraction

11️⃣ Feature extraction helps to:

A) Increase data storage capacity

B) Capture relevant information from input data
C) Remove labels from data
D) Train multiple models simultaneously

✅ Answer: B

12️⃣ Reducing the number of irrelevant variables in a dataset improves:

A) Hardware utilization
B) Model interpretability and performance
C) Data redundancy
D) Data encryption

✅ Answer: B

Supervised Algorithms

13️⃣ Linear regression is mainly used for:

A) Clustering tasks
B) Predicting categorical outcomes
C) Predicting continuous values
D) Dimensionality reduction

✅ Answer: C

14️⃣ Logistic regression is primarily used for:

A) Time series forecasting

B) Binary classification problems
C) Clustering data points
D) Feature extraction

✅ Answer: B

15️⃣ Decision trees make predictions by:

A) Calculating probabilities only

B) Splitting data using feature-based decisions
C) Mapping data into lower dimensions
D) Using distance metrics

✅ Answer: B
16️⃣ Random Forest improves prediction accuracy by:

A) Using a single optimized tree

B) Combining multiple decision trees
C) Reducing dataset size
D) Eliminating features

✅ Answer: B

17️⃣ The primary goal of Support Vector Machines is to:

A) Reduce feature dimensions

B) Maximize margin between classes
C) Identify association rules
D) Cluster similar instances

✅ Answer: B

18️⃣ Naive Bayes classifier assumes that:

A) Features are highly dependent

B) Features are independent
C) Data must be continuous
D) Data must be labeled manually

✅ Answer: B

19️⃣ K-Nearest Neighbors classifies a new instance based on:

A) Probability distributions
B) Distance to nearest data points
C) Decision boundaries
D) Gradient descent optimization

✅ Answer: B
20️⃣ Neural networks are inspired by:

A) Human brain neurons

B) Database structures
C) Statistical distributions
D) Graph theory

✅ Answer: A

21️⃣ Gradient Boosting algorithms improve models by:

A) Training models independently

B) Correcting errors from previous models
C) Removing weak classifiers
D) Reducing dataset size

✅ Answer: B

Applications of Supervised Learning

22️⃣ Predicting whether a borrower will default on a loan is an example of:

A) Clustering
B) Classification
C) Regression
D) Dimensionality reduction

✅ Answer: B

23️⃣ Image recognition systems mainly perform:

A) Classification tasks
B) Association rule mining
C) Clustering tasks
D) Reinforcement learning

✅ Answer: A

24️⃣ Sentiment analysis determines:

A) Emotional tone in text

B) Data structure types
C) Network connections
D) Data compression

✅ Answer: A

Advantages & Limitations

25️⃣ The biggest limitation of supervised learning is:

A) Lack of algorithms
B) Requirement for labeled datasets
C) Limited computational power
D) Inability to generalize

✅ Answer: B

26️⃣ Overfitting occurs when:

A) Model performs well on test data

B) Model memorizes training data
C) Model fails to train properly
D) Model uses fewer features

✅ Answer: B
Unsupervised Learning

27️⃣ Unsupervised learning works primarily with:

A) Labeled data
B) Unlabeled data
C) Reinforcement signals
D) Training labels only

✅ Answer: B

28️⃣ Which task groups similar data points together?

A) Classification
B) Clustering
C) Regression
D) Prediction

✅ Answer: B

29️⃣ PCA is commonly used for:

A) Clustering
B) Classification
C) Dimensionality reduction
D) Reinforcement learning

✅ Answer: C

30️⃣ DBSCAN clustering identifies:

A) Dense clusters and noise points

B) Only spherical clusters
C) Only hierarchical clusters
D) Only numeric datasets
✅ Answer: A

31️⃣ Market basket analysis is an example of:

A) Regression
B) Association rule mining
C) Classification
D) Dimensionality reduction

✅ Answer: B

Reinforcement Learning

32️⃣ Reinforcement learning agents learn primarily through:

A) Labeled datasets
B) Trial and error interactions
C) Clustering techniques
D) Dimensionality reduction

✅ Answer: B

33️⃣ In reinforcement learning, the entity that makes decisions is called:

A) Node
B) Agent
C) Feature
D) Cluster

✅ Answer: B

34️⃣ The environment refers to:

A) Data preprocessing pipeline
B) External system interacting with the agent
C) Training dataset
D) Algorithm parameters

✅ Answer: B

35️⃣ Positive signals given to the agent for good actions are called:

A) Features
B) Rewards
C) States
D) Policies

✅ Answer: B

36️⃣ Markov Decision Processes assume that:

A) Future depends only on past states

B) Future depends only on present state
C) Future depends on all previous states
D) Future cannot be predicted

✅ Answer: B

37️⃣ A policy in reinforcement learning defines:

A) Dataset structure
B) Mapping from states to actions
C) Model training process
D) Reward distribution

✅ Answer: B
38️⃣ Exploration refers to:

A) Using known strategies for rewards

B) Trying new actions to discover better strategies
C) Reducing training data
D) Selecting the best algorithm

✅ Answer: B

39️⃣ Q-learning is a:

A) Model-based algorithm
B) Model-free reinforcement learning algorithm
C) Clustering algorithm
D) Classification algorithm

✅ Answer: B

40️⃣ Deep Q-Networks combine reinforcement learning with:

A) Neural networks
B) Decision trees
C) Logistic regression
D) K-means clustering

✅ Answer: A

MCQs – Meta-Learning, Multi-Agent RL, Scikit-learn

1️⃣ Meta-learning is commonly described as:

A) Learning from labeled datasets only
B) Learning optimal reward functions
C) Learning how to learn from previous experiences
D) Learning through environmental rewards

✅ Answer: C

2️⃣ The main objective of meta-learning is to:

A) Reduce dataset size
B) Improve algorithm efficiency in new tasks
C) Eliminate training data
D) Replace reinforcement learning

✅ Answer: B

3️⃣ In reinforcement learning, meta-learning helps agents to:

A) Store more data in memory
B) Learn policies faster in new environments
C) Avoid interaction with environments
D) Replace exploration strategies

✅ Answer: B

4️⃣ The phrase “learning to learn” is associated with:

A) Deep learning
B) Meta-learning
C) Transfer learning
D) Supervised learning

✅ Answer: B

5️⃣ Meta-learning improves an agent’s ability to:

A) Memorize training datasets
B) Adapt quickly to unseen tasks
C) Reduce neural network size
D) Increase data labeling accuracy

✅ Answer: B

Multi-Agent Reinforcement Learning

6️⃣ Multi-agent reinforcement learning involves:

A) One agent interacting with multiple environments
B) Multiple agents learning and interacting simultaneously
C) One agent using multiple algorithms
D) Multiple datasets used by one agent

✅ Answer: B

7️⃣ In multi-agent RL, agents may:

A) Only cooperate
B) Only compete
C) Cooperate, compete, or communicate
D) Work independently without interaction

✅ Answer: C

8️⃣ Which scenario best represents multi-agent reinforcement learning?

A) A robot navigating a maze alone
B) Multiple robots coordinating in a warehouse
C) A classifier predicting spam emails
D) A regression model predicting prices

✅ Answer: B

9️⃣ A major challenge in multi-agent systems is:

A) Lack of computing power
B) Coordination and cooperation between agents
C) Absence of training data
D) Feature selection

✅ Answer: B

🔟 Multi-agent reinforcement learning is commonly applied in:

A) Image compression
B) Multiplayer game AI
C) Database indexing
D) Data encryption

✅ Answer: B

11️⃣ Economic market simulations using RL involve:

A) Single agent decision making
B) Multiple interacting agents representing participants
C) Unsupervised clustering algorithms
D) Supervised regression models

✅ Answer: B

12️⃣ The emergence of collective behavior in multi-agent systems refers to:

A) Independent agent decisions
B) Agents acting randomly
C) Complex behaviors emerging from agent interactions
D) Agents following identical policies

✅ Answer: C

Scikit-learn
13️⃣ Scikit-learn is primarily known as:
A) A deep learning framework
B) A Python machine learning library
C) A database management system
D) A reinforcement learning simulator

✅ Answer: B

14️⃣ Scikit-learn is widely used because it provides:

A) Hardware acceleration tools
B) Ready-to-use machine learning algorithms
C) Operating system utilities
D) Networking protocols

✅ Answer: B

15️⃣ Which of the following learning tasks is supported by Scikit-learn?

A) Classification
B) Clustering
C) Regression
D) All of these

✅ Answer: D

16️⃣ Dimensionality reduction techniques available in Scikit-learn include:

A) PCA
B) Logistic regression
C) Random forest
D) Gradient boosting

✅ Answer: A

17️⃣ Which task predicts continuous values in Scikit-learn?

A) Regression
B) Clustering
C) Classification
D) Dimensionality reduction

✅ Answer: A

18️⃣ Which learning task assigns labels to data points?

A) Regression
B) Clustering
C) Classification
D) Dimensionality reduction

✅ Answer: C

19️⃣ Which of the following tasks groups similar data points together?
A) Regression
B) Classification
C) Clustering
D) Reinforcement learning

✅ Answer: C

20️⃣ Dimensionality reduction techniques mainly help to:

A) Increase dataset size
B) Reduce number of features while preserving information
C) Replace classification models
D) Eliminate machine learning algorithms

✅ Answer: B