24SJPCCSL508 - MACHINE LEARNING LAB

(Common to CS/CA)

Course Plan
Course Objectives:
To give the learner a practical experience of the various machine learning techniques
and be able to demonstrate them using a language of choice.

Course Outcomes (COs)

• CO1: Understand complexity of Machine Learning algorithms and their limitations.

• CO2: Understand modern notions in data analysis-oriented computing.
• CO3: Apply common Machine Learning algorithms in practice and implement their
own.
• CO4: Perform experiments in Machine Learning using real-world data.

Continuous Assessment (25 Marks)

1. Preparation and Pre-Lab Work (7 Marks)

• Pre-Lab Assignments: Assessment of pre-lab assignments or quizzes that test
understanding of the upcoming experiment.
• Understanding of Theory: Evaluation based on students’ preparation and
understanding of the theoretical background related to the experiments.
2. Conduct of Experiments (7 Marks)
• Procedure and Execution: Adherence to correct procedures, accurate execution of
experiments, and following safety protocols.
• Skill Proficiency: Proficiency in handling equipment, accuracy in observations, and
troubleshooting skills during the experiments.
• Teamwork: Collaboration and participation in group experiments.
3. Lab Reports and Record Keeping (6 Marks)
• Quality of Reports: Clarity, completeness and accuracy of lab reports. Proper
documentation of experiments, data analysis and conclusions.
• Timely Submission: Adhering to deadlines for submitting lab reports/rough record
and maintaining a well-organized fair record.
4. Viva Voce (5 Marks)
• Oral Examination: Ability to explain the experiment, results and underlying principles
during a viva voce session.
Final Marks Averaging: The final marks for preparation, conduct of experiments, viva, and
record are the average of all the specified experiments in the syllabus.

Page 1 of 7
Evaluation Pattern for End Semester Examination (50 Marks)
1. Procedure/Preliminary Work/Design/Algorithm (10 Marks)
• Procedure Understanding and Description: Clarity in explaining the procedure and
understanding each step involved.
• PreliminaryWork and Planning: Thoroughness in planning and organizing
materials/equipment.
• Algorithm Development: Correctness and efficiency of the algorithm related to the
experiment.
• Creativity and logic in algorithm or experimental design.
2. Conduct of Experiment/Execution of Work/Programming (15 Marks)
• Setup and Execution: Proper setup and accurate execution of the experiment or
programming task.
3. Result with Valid Inference/Quality of Output (10 Marks)
• Accuracy of Results: Precision and correctness of the obtained results.
• Analysis and Interpretation: Validity of inferences drawn from the experiment or
quality of program output.
4. Viva Voce (10 Marks)
• Ability to explain the experiment, procedure results and answer related questions
• Proficiency in answering questions related to theoretical and practical aspects of the
subject.
5. Record (5 Marks)
• Completeness, clarity, and accuracy of the lab record submitted

Week Experiment Duration CO(s) Analysis Questions

1. Why does Gradient Descent require
Lab Orientation + CO1,
feature scaling? 2. Compare Gradient
Linear Regression CO2,
1 3 Hours Descent with the Normal Equation. 3. What
(California Housing CO3,
happens if the learning rate is very high or
Dataset) CO4
very low?
CO1, 1. At what polynomial degree does
Polynomial
CO2, overfitting begin? 2. Compare Ridge and
2 Regression + Ridge 3 Hours
CO3, LASSO regularization. 3. Which model
& LASSO Regression
CO4 generalizes better and why?
CO1, 1. Compare MLE and MAP estimation. 2.
Logistic Regression CO2, How does regularization influence parameter
3 3 Hours
using MLE & MAP CO3, estimation? 3. When should MAP be
CO4 preferred over MLE?
Naïve Bayes CO1, 1. Why is Naïve Bayes effective for text
4 3 Hours
Classifier CO2, classification? 2. Compare Multinomial and
Page 2 of 7
Week Experiment Duration CO(s) Analysis Questions
CO3, Bernoulli Naïve Bayes. 3. What happens if
CO4 feature independence is violated?
CO1, 1. How does the value of K affect bias and
K-Nearest Neighbour CO2, variance? 2. Why is feature scaling important
5 3 Hours
(KNN) CO3, in KNN? 3. Compare Euclidean and
CO4 Manhattan distances.
CO1, 1. How do Entropy and Information Gain
CO2, influence tree construction? 2. Why do
6 Decision Tree (ID3) 3 Hours
CO3, Decision Trees overfit? 3. How can pruning
CO4 improve performance?
CO1, 1. Compare Linear, Polynomial and RBF
Support Vector
CO2, kernels. 2. Why is the Maximum Margin
7 Machine (Linear & 3 Hours
CO3, important? 3. When should kernel methods
Kernel SVM)
CO4 be preferred?
CO1, 1. How does the number of hidden layers
Multilayer
CO2, affect performance? 2. Compare Sigmoid,
8 Perceptron (Neural 3 Hours
CO3, ReLU and Tanh activation functions. 3. Why
Network)
CO4 does ReLU often converge faster?
1. Compare Hierarchical Clustering and K-
Hierarchical CO2,
Means. 2. How do you determine the optimal
9 Clustering & K- 3 Hours CO3,
number of clusters? 3. Interpret Silhouette
Means CO4
Score and Inertia.
CO1, 1. Compare Bagging and Boosting. 2. Why
Bagging & Boosting CO2, does boosting reduce bias while bagging
10 3 Hours
Ensemble Methods CO3, reduces variance? 3. Which technique is
CO4 more sensitive to noisy data?
1. Identify underfitting and overfitting from
Bias–Variance CO1,
the generated plots. 2. Explain the
Trade-off using CO2,
11 3 Hours relationship between model complexity and
Polynomial CO3,
generalization. 3. How can regularization
Regression CO4
reduce variance?
CO1,
Practical implementation, viva voce,
Internal Practical CO2,
12 3 Hours debugging, algorithm selection, performance
Test & Viva CO3,
evaluation, and interpretation of results.
CO4

Suggested Deliverables for Every Lab

Each student (or team) should submit:
1. Problem Statement

Page 3 of 7
 2. Algorithm / Flowchart
3. Dataset Description
4. Python Implementation
5. Output Screenshots
6. Performance Metrics
7. Analysis Questions (Mandatory)
8. Conclusion
9. GitHub Repository Link (optional but recommended)

CO-PO-PSO Mapping and Justifications

CO1: Understand complexity of Machine Learning algorithms and their limitations.

CO1 – PO Justifications
PO Value Justification
Understanding the concepts, working principles, and limitations of Machine
Learning algorithms extensively relies on mathematics, statistics, and
PO1 3
computing fundamentals. Since engineering knowledge forms the primary
focus of this CO, the correlation is High (3).
Students analyze the behaviour, assumptions, and limitations of different
PO2 2 Machine Learning algorithms. Problem analysis significantly supports the CO
but is not its primary objective, resulting in a Medium (2) correlation.
Students investigate algorithm behaviour through experiments and compare
PO4 2 outcomes. Investigation supports understanding of the algorithms but is not
the dominant activity, leading to a Medium (2) correlation.
Students use Machine Learning tools and programming libraries to study
PO5 2 algorithm performance. Tool usage is essential for learning but supports
rather than drives the CO, resulting in a Medium (2) correlation.
The CO introduces modern Machine Learning concepts that encourage
PO11 2 students to keep pace with evolving AI technologies. Lifelong learning is
promoted but not the principal outcome, giving a Medium (2) correlation.

CO2: Understand modern notions in data analysis-oriented computing.

CO2 – PO Justifications
PO Value Justification
Understanding data analysis techniques requires extensive application of
PO1 3 engineering, statistical, and computing fundamentals. Engineering knowledge
is the core requirement of this CO, resulting in a High (3) correlation.
Students examine datasets, identify patterns, and determine suitable
PO2 3 analytical techniques. Since problem analysis is the primary learning activity,
the correlation is High (3).

Page 4 of 7
PO Value Justification
Students investigate datasets through experimentation and interpret analytical
PO4 2 results. Investigation supports data analysis but is not the primary emphasis,
resulting in a Medium (2) correlation.
Students extensively use Python, Scikit-learn, Pandas, and visualization tools
PO5 3 for data analysis. Modern tool usage is integral to achieving this CO, giving a
High (3) correlation.
Exposure to modern data analytics techniques encourages students to learn
PO11 2 emerging technologies independently. Lifelong learning is supported but is not
the principal objective, resulting in a Medium (2) correlation.

CO3: Apply common Machine Learning algorithms in practice and implement their own.
CO3 – PO Justifications
PO Value Justification
Implementing Machine Learning algorithms extensively applies engineering,
PO1 3 mathematical, and computing knowledge. Engineering knowledge is a core
component of this CO, resulting in a High (3) correlation.
Students analyze practical problems, select suitable algorithms, and interpret
PO2 3 implementation results. Problem analysis is central to achieving this CO, giving
a High (3) correlation.
Students develop Machine Learning solutions for practical applications.
PO3 2 Solution development is significantly addressed, but implementation rather
than design is the main emphasis, resulting in a Medium (2) correlation.
Students perform experiments, validate algorithms, and compare model
PO4 3 performance using real datasets. Investigation is a major laboratory activity,
resulting in a High (3) correlation.
Students extensively implement algorithms using modern Machine Learning
PO5 3 frameworks and development tools. Tool usage is essential for accomplishing
this CO, giving a High (3) correlation.
Students collaborate during implementation, debugging, and discussions.
PO8 2 Teamwork supports laboratory activities but is not the primary learning
outcome, resulting in a Medium (2) correlation.
Students communicate implementation results through laboratory records and
PO9 1 viva voce. Communication is only indirectly addressed as a supporting activity;
therefore, the correlation is Low (1).
Students gain experience with multiple Machine Learning techniques,
encouraging independent exploration of advanced methods. Lifelong learning
PO11 2
is supported but is not the principal focus, resulting in a Medium (2)
correlation.

Page 5 of 7
CO4: Perform experiments in Machine Learning using real-world data.
CO4 – PO Justifications
PO Value Justification
Students apply engineering and computing fundamentals while conducting
PO1 2 experiments. Engineering knowledge supports experimental work but is not
the primary objective, resulting in a Medium (2) correlation.
Students analyze real-world datasets, compare algorithm performance, and
PO2 3 interpret outcomes. Problem analysis is the primary activity throughout the
experiments, giving a High (3) correlation.
Students modify and apply Machine Learning models to solve different
PO3 2 practical problems. Solution development is significantly supported but is not
the dominant activity, resulting in a Medium (2) correlation.
Designing experiments, validating results, and drawing conclusions are the
PO4 3
core objectives of this CO. Hence, the contribution to investigation is High (3).
Students extensively use Machine Learning libraries, visualization software,
PO5 3 and development tools throughout the laboratory. Tool usage is fundamental
to achieving this CO, resulting in a High (3) correlation.
Students perform experiments collaboratively, share responsibilities, and
PO8 2 participate in technical discussions. Teamwork supports the laboratory but is
not the principal focus, resulting in a Medium (2) correlation.
Students prepare laboratory reports, explain observations, and present
PO9 2 findings during viva. Communication is significantly developed but remains a
supporting outcome, resulting in a Medium (2) correlation.
Experimental work encourages students to independently explore new
datasets and emerging Machine Learning techniques. Lifelong learning is
PO11 2
supported but is not the primary objective, resulting in a Medium (2)
correlation.

CO–PSO Justifications
CO1
PSO Value Justification
Understanding Machine Learning algorithms forms the foundation for
PSO1 3 analyzing and developing computing solutions. This is the primary objective
of the CO, resulting in a High (3) correlation.
The CO introduces Machine Learning as an emerging ICT technology.
PSO3 2 Adaptation to new technologies is supported, but innovation is not the
primary emphasis, resulting in a Medium (2) correlation.

CO2

Page 6 of 7
PSO Value Justification
Data analysis concepts directly strengthen students' ability to analyze and
PSO1 3 develop computing solutions. This is a core outcome of the CO, resulting in a
High (3) correlation.
Students implement data analysis techniques using software tools and
PSO2 2 structured programming practices. Software engineering supports the CO but
is not the primary focus, resulting in a Medium (2) correlation.
Modern data analysis and Machine Learning are key emerging ICT
PSO3 3 technologies. Developing competency in these areas is a primary objective of
the CO, resulting in a High (3) correlation.

CO3
PSO Value Justification
Implementing Machine Learning algorithms directly develops computing
PSO1 3 solutions using Computer Science fundamentals. This is the core activity of the
CO, resulting in a High (3) correlation.
Students design, implement, test, and validate Machine Learning applications
PSO2 3 using sound software engineering practices. Software development is a
primary outcome, resulting in a High (3) correlation.
Students implement advanced Machine Learning algorithms widely used in
PSO3 3 emerging AI applications. Adapting to emerging technologies is a major
objective of this CO, giving a High (3) correlation.

CO4
PSO Value Justification
Conducting experiments on real-world datasets directly strengthens students'
PSO1 3 ability to develop and evaluate computing solutions. This forms the primary
objective of the CO, resulting in a High (3) correlation.
Students implement, test, validate, and document Machine Learning
PSO2 3 applications using systematic software engineering practices. These activities
are central to the laboratory, resulting in a High (3) correlation.
Working with real-world datasets and modern Machine Learning frameworks
PSO3 3 develops competency in emerging AI and ICT technologies. This is a core
outcome of the CO, resulting in a High (3) correlation.

Page 7 of 7