Fundamentals of Machine Learning (4341603) - Summer 2025 Solution

Question 1(a) [3 marks]

#

Define machine Learning. Give any two applications of machine learning.

Answer:

Machine Learning is a subset of artificial intelligence that enables computers to learn and make decisions from data without being explicitly programmed for every task.

Applications:

• Email spam detection: Automatically identifies and filters spam emails
• Recommendation systems: Suggests products on e-commerce sites like Amazon

Table: ML vs Traditional Programming

Mnemonic: “ML = Make Learning from data”

Question 1(b) [4 marks]

#

Define: Under fitting and overfitting.

Answer:

Underfitting occurs when a model is too simple to capture underlying patterns in data, resulting in poor performance on both training and test data.

Overfitting occurs when a model learns training data too well, including noise, causing poor performance on new unseen data.

Table: Comparison

Mnemonic: “Under = Under-performs, Over = Over-learns”

Question 1(c) [7 marks]

#

Describe different types of machine learning with suitable example.

Answer:

Table: Types of Machine Learning

Supervised Learning uses input-output pairs to train models. The algorithm learns from examples to predict outcomes for new data.

Unsupervised Learning discovers hidden patterns in data without target labels. It groups similar data points together.

Reinforcement Learning trains agents to make decisions by rewarding good actions and penalizing bad ones.

Diagram:

graph TD
    A[Machine Learning] --> B[Supervised Learning]
    A --> C[Unsupervised Learning]
    A --> D[Reinforcement Learning]
    B --> E[Classification]
    B --> F[Regression]
    C --> G[Clustering]
    C --> H[Association Rules]

Mnemonic: “Super Un-supervised Reinforces learning”

Question 1(c) OR [7 marks]

#

Describe different tools and technology used in the field machine learning.

Answer:

Table: ML Tools and Technologies

Python is the most popular language due to its simplicity and extensive libraries.

Scikit-learn provides simple tools for data mining and analysis, perfect for beginners.

TensorFlow and PyTorch are advanced frameworks for deep learning applications.

Jupyter Notebook offers interactive development environment for experimentation.

Diagram:

graph LR
    A[Data] --> B[Pandas/NumPy]
    B --> C[Scikit-learn]
    C --> D[Model]
    D --> E[Matplotlib]

Mnemonic: “Python Pandas Scikit Tensor Jupyter”

Question 2(a) [3 marks]

#

Give the difference between Qualitative data and Quantitative data.

Answer:

Table: Qualitative vs Quantitative Data

Qualitative data describes qualities or characteristics that cannot be measured numerically.

Quantitative data represents measurable quantities expressed as numbers.

Mnemonic: “Quality = Categories, Quantity = Numbers”

Question 2(b) [4 marks]

#

Find the mean and median for the following data: 3,4,5,5,7,8,9,11,12,14.

Answer:

Given data: 3, 4, 5, 5, 7, 8, 9, 11, 12, 14

Mean calculation:

• Sum = 3+4+5+5+7+8+9+11+12+14 = 78
• Count = 10 numbers
• Mean = 78/10 = 7.8

Median calculation:

• Data is already sorted
• For 10 numbers: Median = (5th + 6th value)/2
• Median = (7+8)/2 = 7.5

Table: Results

Mnemonic: “Mean = Average, Median = Middle”

Question 2(c) [7 marks]

#

Describe machine learning activities in detail.

Answer:

Table: Machine Learning Activities

Data Collection involves gathering information from various sources like databases, sensors, or surveys.

Data Preprocessing includes cleaning, transforming, and organizing raw data for analysis.

Feature Selection identifies the most relevant variables that contribute to predictions.

Model Training uses algorithms to learn patterns from prepared training data.

Model Evaluation tests how well the trained model performs on new, unseen data.

Diagram:

flowchart TD
    A[Data Collection] --> B[Data Preprocessing]
    B --> C[Feature Selection]
    C --> D[Model Training]
    D --> E[Model Evaluation]
    E --> F[Deployment]

Mnemonic: “Collect Process Feature Train Evaluate Deploy”

Question 2(a) OR [3 marks]

#

Give the difference between predicative model and descriptive model.

Answer:

Table: Predictive vs Descriptive Models

Predictive models use historical data to make predictions about future events or unknown outcomes.

Descriptive models analyze existing data to understand current patterns and relationships.

Mnemonic: “Predict = Future, Describe = Present”

Question 2(b) OR [4 marks]

#

Classify the following using appropriate data type: hair color, gender, blood group type, time of day.

Answer:

Table: Data Type Classification

Nominal data represents categories without any natural ordering.

Continuous data can take any value within a range and is measurable.

Mnemonic: “Names = Nominal, Numbers = Numerical”

Question 2(c) OR [7 marks]

#

Explain various methods used in data pre-processing.

Answer:

Table: Data Preprocessing Methods

Data Cleaning removes or corrects erroneous, incomplete, or irrelevant data.

Data Integration combines data from multiple sources into a unified dataset.

Data Transformation converts data into appropriate formats for analysis.

Data Reduction decreases dataset size while maintaining information quality.

Diagram:

graph TD
    A[Raw Data] --> B[Data Cleaning]
    B --> C[Data Integration]
    C --> D[Data Transformation]
    D --> E[Data Reduction]
    E --> F[Processed Data]

Mnemonic: “Clean Integrate Transform Reduce”

Question 3(a) [3 marks]

#

Give difference between classification and regression.

Answer:

Table: Classification vs Regression

Classification predicts discrete categories or classes from input data.

Regression predicts continuous numerical values from input data.

Mnemonic: “Class = Categories, Regress = Real numbers”

Question 3(b) [4 marks]

#

Write confusion matrix using appropriate example. Calculate accuracy and error rate for it.

Answer:

Example: Email Classification

Table: Confusion Matrix

Calculations:

• Accuracy = (TP+TN)/(TP+TN+FP+FN) = (85+90)/200 = 87.5%
• Error Rate = (FP+FN)/(TP+TN+FP+FN) = (10+15)/200 = 12.5%

Key Terms:

• TP: True Positive - Correctly predicted spam
• TN: True Negative - Correctly predicted not spam

Mnemonic: “True Positive True Negative = Correct predictions”

Question 3(c) [7 marks]

#

Explain KNN algorithm in detail.

Answer:

K-Nearest Neighbors (KNN) is a simple classification algorithm that classifies data points based on the majority class of their K nearest neighbors.

Table: KNN Algorithm Steps

Working Process:

1. Calculate distances between test point and all training points
2. Sort distances and select K nearest neighbors
3. Count votes from each class among neighbors
4. Assign class with majority votes

Diagram:

graph TD
    A[New Data Point] --> B[Calculate Distances]
    B --> C[Find K Nearest]
    C --> D[Majority Vote]
    D --> E[Predict Class]

Advantages:

• Simple to implement and understand
• No training required - lazy learning algorithm

Mnemonic: “K Nearest Neighbors Vote for classification”

Question 3(a) OR [3 marks]

#

Give any three applications of multiple linear regression.

Answer:

Applications of Multiple Linear Regression:

Table: Applications

Multiple Linear Regression uses multiple input variables to predict a continuous output variable.

Mnemonic: “Multiple inputs, One output”

Question 3(b) OR [4 marks]

#

Explain bagging, boosting and stacking in detail.

Answer:

Table: Ensemble Methods

Bagging trains multiple models on different data subsets and averages predictions.

Boosting trains models sequentially, each learning from previous model’s mistakes.

Stacking uses a meta-model to learn how to combine predictions from base models.

Mnemonic: “Bag parallel, Boost sequential, Stack meta”

Question 3(c) OR [7 marks]

#

Explain single linear regression with its application.

Answer:

Single Linear Regression finds the best straight line relationship between one input variable (X) and one output variable (Y).

Formula: Y = a + bX

• a: Y-intercept
• b: Slope of line

Table: Application Example - House Price vs Size

Working Process:

1. Collect data with input-output pairs
2. Plot points on scatter graph
3. Find best line that minimizes error
4. Make predictions using line equation

Diagram:

Applications:

• Sales vs Advertising: More ads → More sales
• Temperature vs Ice cream sales: Hot weather → More sales

Mnemonic: “One X predicts One Y with a line”

Question 4(a) [3 marks]

#

Define the following: (1)support (2) confidence.

Answer:

Support measures how frequently an itemset appears in the dataset.

Confidence measures how often items in consequent appear when antecedent is present.

Table: Definitions

Support = Frequency of occurrence Confidence = Reliability of rule

Mnemonic: “Support = How often, Confidence = How reliable”

Question 4(b) [4 marks]

#

Explain applications of unsupervised learning.

Answer:

Table: Unsupervised Learning Applications

Customer Segmentation groups customers with similar buying behavior for targeted marketing.

Data Compression reduces storage space by finding patterns and removing redundancy.

Anomaly Detection identifies unusual patterns that may indicate fraud or errors.

Mnemonic: “Segment Compress Detect Recommend”

Question 4(c) [7 marks]

#

Write and explain apriori algorithm with suitable example.

Answer:

Apriori Algorithm finds frequent itemsets and generates association rules for market basket analysis.

Table: Algorithm Steps

Example Dataset:

• Transaction 1: {Bread, Milk, Eggs}
• Transaction 2: {Bread, Milk}
• Transaction 3: {Bread, Eggs}
• Transaction 4: {Milk, Eggs}

Working Process:

1. Scan database to count item frequencies
2. Generate candidate itemsets of increasing size
3. Prune infrequent itemsets below minimum support
4. Generate association rules from frequent itemsets

Diagram:

flowchart TD
    A[Transaction Database] --> B[Find Frequent 1-itemsets]
    B --> C[Generate 2-itemsets]
    C --> D[Apply Min Support]
    D --> E[Generate Rules]

Mnemonic: “A-priori knowledge helps find frequent patterns”

Question 4(a) OR [3 marks]

#

List out the difference between clustering and classification.

Answer:

Table: Clustering vs Classification

Clustering discovers hidden groups in unlabeled data.

Classification assigns new data to known categories using trained models.

Mnemonic: “Cluster = Groups unknown, Classify = Labels known”

Question 4(b) OR [4 marks]

#

Explain the clustering process in detail.

Answer:

Table: Clustering Process Steps

Clustering Process groups similar data points together based on their characteristics.

Key decisions include choosing the number of clusters and appropriate distance metrics.

Validation ensures clusters are meaningful and well-separated.

Mnemonic: “Prepare Distance Algorithm Validate”

Question 4(c) OR [7 marks]

#

Write and explain K-means clustering algorithm with suitable example.

Answer:

K-means partitions data into K clusters by minimizing within-cluster sum of squares.

Table: Algorithm Steps

Example: Customer Income vs Age

• Customer 1: (Income=30k, Age=25)
• Customer 2: (Income=35k, Age=30)
• Customer 3: (Income=70k, Age=45)
• Customer 4: (Income=75k, Age=50)

Working Process:

1. Choose K=2 clusters for young/old customers
2. Initialize centroids randomly
3. Calculate distances from each customer to centroids
4. Assign customers to nearest centroid
5. Update centroid positions to center of assigned customers
6. Repeat until stable

Diagram:

flowchart TD
    A[Choose K] --> B[Initialize Centroids]
    B --> C[Assign Points to Nearest Centroid]
    C --> D[Update Centroid Positions]
    D --> E{Converged?}
    E -->|No| C
    E -->|Yes| F[Final Clusters]

Advantages:

• Simple and fast for large datasets
• Works well with spherical clusters

Mnemonic: “K centroids Mean their assigned points”

Question 5(a) [3 marks]

#

List the applications of matplotlib.

Answer:

Table: Matplotlib Applications

Matplotlib is Python’s primary plotting library for creating static, animated, and interactive visualizations.

Key features include support for multiple plot types and customizable styling.

Mnemonic: “Mat-plot-lib = Math Plotting Library”

Question 5(b) [4 marks]

#

Write down code to plot a vertical line and horizontal line using matplotlib.

Answer:

Code Block:

import matplotlib.pyplot as plt

# Create figure
plt.figure(figsize=(8, 6))

# Plot vertical line at x=3
plt.axvline(x=3, color='red', linestyle='--', label='Vertical Line')

# Plot horizontal line at y=2
plt.axhline(y=2, color='blue', linestyle='-', label='Horizontal Line')

# Add labels and title
plt.xlabel('X-axis')
plt.ylabel('Y-axis')
plt.title('Vertical and Horizontal Lines')
plt.legend()
plt.grid(True)
plt.show()

Key Functions:

• axvline(): Creates vertical line
• axhline(): Creates horizontal line

Mnemonic: “axvline = Vertical, axhline = Horizontal”

Question 5(c) [7 marks]

#

Explain features and applications of Scikit-Learn.

Answer:

Table: Scikit-Learn Features

Scikit-Learn is Python’s most popular machine learning library providing simple tools for data analysis.

Key Strengths:

• Consistent interface across all algorithms
• Excellent documentation with examples
• Active community support and development

Applications:

• Classification: Email spam detection
• Regression: House price prediction
• Clustering: Customer segmentation
• Dimensionality Reduction: Data visualization

Diagram:

graph TD
    A[Scikit-Learn] --> B[Classification]
    A --> C[Regression]
    A --> D[Clustering]
    A --> E[Preprocessing]
    B --> F[SVM, Decision Trees]
    C --> G[Linear, Polynomial]
    D --> H[K-means, DBSCAN]
    E --> I[Scaling, Encoding]

Mnemonic: “Scikit = Science Kit for machine learning”

Question 5(a) OR [3 marks]

#

Give the purpose of NumPy in machine learning.

Answer:

Table: NumPy Purpose in ML

NumPy provides the foundation for numerical computing in Python machine learning applications.

Essential for handling large datasets and performing mathematical operations efficiently.

Mnemonic: “Num-Py = Numerical Python”

Question 5(b) OR [4 marks]

#

Write down steps to import csv file in pandas.

Answer:

Code Block:

import pandas as pd

# Step 1: Import pandas library
# Step 2: Use read_csv() function
data = pd.read_csv('filename.csv')

# Step 3: Display first few rows
print(data.head())

# Optional: Specify parameters
data = pd.read_csv('file.csv', 
                   delimiter=',',
                   header=0,
                   index_col=0)

Steps:

1. Import pandas library
2. Use read_csv() function with filename
3. Verify data with head() method

Mnemonic: “Import Read Verify”

Question 5(c) OR [7 marks]

#

Explain features and applications of Pandas.

Answer:

Table: Pandas Features

Pandas is the primary data manipulation library in Python for machine learning projects.

Key Capabilities:

• Data Loading from various file formats
• Data Cleaning and preprocessing operations
• Data Transformation and reshaping
• Statistical Analysis and aggregation

Applications:

• Data Preprocessing: Clean datasets before ML
• Exploratory Analysis: Understand data patterns
• Feature Engineering: Create new variables
• Data Integration: Merge multiple data sources

Diagram:

graph TD
    A[Raw Data] --> B[Pandas DataFrame]
    B --> C[Data Cleaning]
    C --> D[Data Analysis]
    D --> E[Feature Engineering]
    E --> F[ML Ready Data]

Advantages:

• Intuitive syntax for data operations
• High performance with optimized operations
• Integration with other ML libraries

Mnemonic: “Pandas = Panel Data for analysis”