A

Project Report On

Student Learning Style Prediction (VARK) Using Machine

Learning Algorithms
Submitted in Partial Fulfilment of the Requirements
for the Degree of
Bachelor of Technology
in
Computer Engineering
by
Ms. Ankita Patil (2003097)
Ms. Kartiki Pise (2003103)
Ms. Sanskruti Bhivare (2003105)
Ms. Nikita Patil (2003111)
Under The Guidance Of
Prof. V. S. Mokashi

Department of Computer Engineering

K.E. Society’s
Rajarambapu Institute of Technology, Rajaramnagar
(An Empowered Autonomous Institute, Affiliated to Shivaji University, Kolhapur)
2023-2024
 CERTIFICATE

This is to certify that Ms. Ankita Patil, Ms. Kartiki Pise , Ms. Sanskruti Bhivare and
Ms. Nikita Patil has successfully completed the project work and submitted project report
on “Student Learning Style Prediction (VARK) using Machine Learning Algorithms” for the
partial fulfillment of the requirement for the degree of Bachelor of Technology in Computer
engineering at Rajarambapu Institute of Technology, Rajaramnagar, Dist: Sangli. This final
report is the record of the students work carried out under my supervision and guidance.

Date:
Place: RIT, Rajaramnagar

Prof. V. S. Mokashi Dr. S. U. Mane

Asst. Professor Head of Department

Name and Sign of External Examiner:-

 DECLARATION

We, the undersigned, hereby declare that the project report entitled “Student Learning Style
Prediction (VARK) using Machine Learning Algorithms” written and submitted by us to
Computer Engineering department, under the guidance of Prof. V. S. Mokashi is our original
work. The empirical results in this project report are based on the data collected by us.

Sr.No Student Name Roll No Signature

1 Ankita Anandrao Patil 2003097

2 Kartiki Suresh Pise 2003103

3 Sanskruti Sunil Bhivare 2003105

4 Nikita Suresh Patil 2003111

 ACKNOWLEDGEMENT

It is our foremost duty to express our deep sense of gratitude and respect to the guide
Prof. V. S. Mokashi for his uplifting tendency and inspiring us for taking up this project
work successful.
We are also grateful to Dr. S. U. Mane (Head of Department of Computer Engineer-
ing) for providing all necessary facilities to carry out the project work and whose encouraging
part has been a perpetual source of information.
We are highly indebted to Dr. P. V. Kadole for their guidance and constant supervision
as well as for providing necessary information regarding the project also for their support in
completing the project.
We also thank all staff members of our Department for their timely help and encour-
agement, which help us in completing of our project work.
We are indebted to the library personnel’s for offering all the help in completing the
project work. Last but not only the least we are thankful to our colleagues and those helped
us directly or indirectly throughout this project work.
 ABSTRACT

Effective teaching strategies rely on a comprehensive grasp of students’ learning styles.

The study predicts student’s learning preferences based on the VARK model (Visual, Audi-
tory, Reading/Writing and Kinesthetic modalities) using machine learning algorithms. The
dataset includes a variety of learning preferences that were gathered via questionnaires and
tests.
A variety of machine learning methods, including Decision Trees, Support vector ma-
chines (SVM), K-Nearest Neighbours (KNN), and Neural Networks, are used to build pre-
dictive models.
To evaluate model effectiveness and compare the predictive capabilities of various al-
gorithms, performance metrics like accuracy, precision, recall, and F1-score are employed.
The results demonstrate the efficacy of certain algorithms in predicting specific learning
styles. This research contributes to enhancing educational methodologies by providing in-
sights into predicting students’ learning preferences, facilitating personalized teaching ap-
proaches tailored to individual learning styles.

Keywords: Machine learning, Algorithms, Support Vector Machine, K-Means cluster-

ing, KNN, Neural Networks, VARK, Confusion Matrix.

i
Contents

1 Introduction 1

2 Problem Life Cycle 4

2.1 Problem Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Problem Selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3.1 Objectives:- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.4 Problem Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.4.1 Why Why analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.5 End Users . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3 Literature Survey And Motivation 8

4 Proposed System and Requirement Specification 10

4.1 Proposed System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
4.2 Software Requirements Specification-SRS . . . . . . . . . . . . . . . . . . 11
4.2.1 Modules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4.2.2 Functional Requirements . . . . . . . . . . . . . . . . . . . . . . . 11
4.2.3 Non-Functional Requirements . . . . . . . . . . . . . . . . . . . . 12
4.3 Significance of the project . . . . . . . . . . . . . . . . . . . . . . . . . . 13

ii
 4.4 Scope of Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.5 Deployment Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.5.1 Hardware Requirements . . . . . . . . . . . . . . . . . . . . . . . 16
4.5.2 Software Requirements . . . . . . . . . . . . . . . . . . . . . . . . 17
4.6 Others . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.7 Project Cost Estimate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.8 Project Deliverable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.9 Project Success . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

5 Design 21
5.1 System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.2 Data Flow Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.3 UML Class Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
5.4 Use Case Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

6 Development/Implementation Details 28
6.1 Technology Used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6.1.1 Graphical user interface (GUI) . . . . . . . . . . . . . . . . . . . . 28
6.1.2 Python . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6.1.3 OpenCV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.1.4 Visual Studio Code . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.2 Data Processing and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.2.1 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
6.2.2 Data Pre-processing . . . . . . . . . . . . . . . . . . . . . . . . . 30
6.2.3 Data Splitting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
6.3 Machine Learning Algorithms . . . . . . . . . . . . . . . . . . . . . . . . 31
6.3.1 Supervised Classification algorithms: . . . . . . . . . . . . . . . . 31

iii
 6.3.2 Unsupervised Classification Algorithms: . . . . . . . . . . . . . . 37

7 Testing 41
7.1 Test Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

8 Deployment 44
8.1 Readme File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
8.2 User Manual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8.2.1 Home Page . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
8.2.2 Recommendation Page . . . . . . . . . . . . . . . . . . . . . . . . 45
8.3 GUI for Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

9 Results and Discussion 47

9.1 Classification Report of all algorithms . . . . . . . . . . . . . . . . . . . . 48

10 Conclusion and Future Work 50

10.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
10.2 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

11 References 53

12 Plagiarism Report 56

13 Activity Chart 57

14 QUANTUM 2k23 Poster 58

iv
List of Figures

5.1 System Architecture of Learning style prediction model . . . . . . . . . . . 21

5.2 Data Flow diagram of Learning style prediction model . . . . . . . . . . . 23
5.3 UML Class Diagram of Learning style prediction model . . . . . . . . . . 25
5.4 Use Case Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

6.1 Data collection sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

6.2 Numeric Output after Data Preprocessing . . . . . . . . . . . . . . . . . . 30
6.3 Splitting of Data into training and testing . . . . . . . . . . . . . . . . . . . 31
6.4 Implementation of Logistic Regression . . . . . . . . . . . . . . . . . . . 32
6.5 Implementation of Decision tree . . . . . . . . . . . . . . . . . . . . . . . 32
6.6 Implementation of Random Forest . . . . . . . . . . . . . . . . . . . . . . 33
6.7 Implementation of KNN(k-Nearest Neighbors) . . . . . . . . . . . . . . . 34
6.8 Implementation of Support Vector Machine (SVM) . . . . . . . . . . . . . 35
6.9 Implementation of Naive Bayes . . . . . . . . . . . . . . . . . . . . . . . 36
6.10 Implementation of K-Means Clustering . . . . . . . . . . . . . . . . . . . 37
6.11 Clusters of Customer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
6.12 Hierarchical Clustering Dendrogram . . . . . . . . . . . . . . . . . . . . . 39
6.13 Hierarchical Clustering result . . . . . . . . . . . . . . . . . . . . . . . . . 40

9.1 Result of Supervised Classification Algorithm . . . . . . . . . . . . . . . . 48

v
9.2 Result of Un-Supervised Classification Algorithm . . . . . . . . . . . . . . 48
9.3 Accuracy Comparison of Different Classifiers . . . . . . . . . . . . . . . . 49

13.1 Activity Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

14.1 Poster . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

vi
List of Tables

7.1 Student Learning style based on VARK model . . . . . . . . . . . . . . . . 42

7.2 Final Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

vii
Chapter 1

Introduction

1.1 Motivation for present work

Applying machine learning algorithms for forecasting student learning styles according to
the VARK model yields valuable insights for customizing education. The motivation behind
this effort lies in improving the learning journey, adapting instructional approaches to indi-
vidual preferences, promoting heightened engagement, and ultimately enhancing academic
performance. The objective is to establish a more streamlined and imactful educational sys-
tem that caters to a range of learning styles, resulting in heightened student success and
satisfaction.

1.2 Problem Statement

Student Learning Style Prediction (VARK) using Machine Learning Algorithms.

1.3 Nature inspired algorithms

Consider employing nature-inspired algorithms such as Genetic Algorithm, Particle Swarm
Optimization, Ant Colony Optimization, Firefly Algorithm, and Artificial Bee Colony within
your predictive models for assessing student learning styles through the VARK model with

1
 Student Learning style prediction based on VARK using Machine Learning

machine learning algorithms. These algorithms mimic natural phenomena like genetic evo-
lution, swarm intelligence, ant foraging, firefly flashing patterns, and bee colony optimiza-
tion, aiming to address optimization or classification challenges in predicting learning styles.

1.4 Research Gaps

Limited Dataset Availability: Collecting data through survey form student might be chal-
lenging Feature Selection and Representation: Feature selection is complex and requiring
in- depth analysis to identify the most relevant variables. Algorithm Selection and Per-
formance: Identifying the most suitable machine learning algorithms that effectively learn
from the limited dataset, generalize well, and accurately predict learning styles remains a
challenge. Interpretability and Explainability: For the VARK learning style prediction, the
challenge lies in creating models that not only accurately predict learning styles but also pro-
vide interpretable explanations for why a particular learning style was assigned to a student.
Cross-Cultural Validity: In the context of learning styles, it pertains to whether the VARK
model’s categorization of individuals into visual, auditory, reading/writing, or kinesthetic
learners remains consistent and effective across various cultural backgrounds.

1.5 Layout of Work

1.5.1 Data Collection and Preprocessing:

Provide information on the sources and methodologies employed for gathering student
data, encompassing responses from the VARK questionnaire. Delve into the processes of
data cleansing, addressing missing values, and encoding variables that are categorical in na-
ture.

RIT, Rajaramnagar 2 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

1.5.2 Feature Selection and Engineering:

Highlight the selection of relevant features (V, A, R, K) and any additional parameters
derived from the data for model development.

1.5.3 Machine Learning Algorithms:

Explain the ML algorithms employed (e.g., Decision Trees, Random Forest, SVM) and
the rationale behind their selection. Discuss the model’s training, validation, and evaluation
techniques used.

1.5.4 Model Performance and Results:

Share the accuracy, precision, recall, and F1-score metrics of the created model, offering
insights into its efficacy for predicting learning styles.

1.5.5 Interpretation and Conclusion:

Examine the importance of the model’s predictions within the educational context, ex-
ploring potential effects on teaching methods and personalized learning. Provide a concise
overview of the results and propose directions for additional research.

RIT, Rajaramnagar 3 Dept. of CSE

Chapter 2

Problem Life Cycle

2.1 Problem Identification

A one-size-fits-all strategy frequently fails in modern education to meet the varied learning
demands of individuals. These preferences are categorized by the VARK model (Visual,
Auditory, Reading/Writing, Kinesthetic), however it takes a lot of effort and is prone to error
to manually determine each student’s preferred learning style. The task of applying machine
learning algorithms to predict the learning styles of students is taken on in this capstone
project. The project intends to automate and improve learning style identification accuracy
through input data analysis. In addition to saving teachers time, this guarantees that students
have individualized learning experiences, which raises engagement and improves academic
achievement. The advancement of adaptive learning technology depends on recognizing and
resolving these problems.

4
 Student Learning style prediction based on VARK using Machine Learning

2.2 Problem Selection

The difficulty of properly accommodating varied learning preferences in educational con-

texts is the main emphasis of the project. The research tries to accurately predict students
learning styles by utilizing machine learning techniques and the VARK model. This predic-
tion makes it possible to create individualized learning experiences that improve understand-
ing and engagement. By matching teaching strategies with individual learning preferences,
the project improves educational outcomes by addressing the need for adaptive learning so-
lutions.

2.3 Problem Definition

The study uses the VARK approach (Visual, Auditory, Reading/Writing, Kinesthetic) to
effectively determine pupils’ particular learning preferences. Conventional techniques for
identifying learning styles are frequently arbitrary and ineffective. This project uses machine
learning algorithms to anticipate the preferred learning styles of students based on the data
they provide. This will allow for individualized learning experiences and improve learning
outcomes by using teaching techniques and content delivery that are specifically designed
for each student.

2.3.1 Objectives:-

1. Accurately predicting a learner’s preferred learning style by analyzing their their input
data.

2. To identify patterns and correlations in the training data that can be used to improve
learning style prediction accuracy.

RIT, Rajaramnagar 5 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

3. To classify learners into different learning style categories, including visual, audi-
tory,kinesthetic, and other styles.

4. Offering personalized learning recommendations tailored to a learner’s predicted learn-

ing style, encompassing suitable learning materials, activities, and assessment formats

2.4 Problem Analysis

Understanding the various learning preferences among students is a key component of

the problem analysis for the project on student learning style prediction based on the VARK
model utilizing machine learning. Learning styles are divided into four categories under the
VARK model: visual, auditory, reading/writing, and kinesthetic. To improve student en-
gagement and academic success, educational approaches can be personalized by properly
predicting individual learning styles. The analysis comprises looking at a number of char-
acteristics, including academic credentials, technical skills, and demographic variables, that
influence preferences for learning styles. It also entails evaluating how well conventional
teaching techniques accommodate different learning styles and pinpointing areas in need of
development. The research intends to create a strong machine learning model that can pre-
cisely anticipate students’ learning preferences based on a thorough problem analysis.With
this method, teachers can modify their lesson plans to better meet the needs of each student,
creating an environment that is both inclusive and productive.

2.4.1 Why Why analysis

The "Why-Why Analysis" is a technique used to identify the root causes of a problem by
asking "why" repeatedly. Let’s apply this technique to understand the reasons behind the im-
plementation of a Student learning style prediction based on VARK using Machine learning:
1. Why Examine Learning Styles of Students?

RIT, Rajaramnagar 6 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

- Educational efficacy is increased when individual learning styles are recognized and ac-
commodated. Anticipating different learning patterns facilitates tailored instruction, which
enhances student involvement and scholastic achievement.
2. The VARK Model: Why Use It?
- The VARK model is a well-known paradigm for classifying learning preferences. It stands
for Visual, Auditory, Reading/Writing, Kinesthetic. It offers a thorough grasp of how pupils
take in and process information.
3. Why Use Algorithms for Machine Learning?
- The process of manually identifying learning styles is laborious and prone to bias. Large
datasets can be analyzed quickly by machine learning algorithms, which can also spot pat-
terns and accurately anticipate learning styles. This ensures that the solutions are scalable
and objective.
4. Why Is Accurate Prediction So Important?
- Precise forecasts guarantee that educational materials correspond with students’ require-
ments, resulting in enhanced learning achievements. There may be misalignment.

2.5 End Users

Curriculum creators, educational institutions, and educators are the end users of the
"Student Learning Style Prediction based on VARK Model using Machine Learning" project.
Teachers gain from knowing specific student preferences since it makes it easier for them
to modify their teaching strategies. These insights are used by curriculum creators to create
content that suits a variety of learning styles and improves understanding and overall engage-
ment. This method is used by educational institutions to increase learning outcomes, tailor
instruction at scale, and create a more welcoming and productive learning environment.

RIT, Rajaramnagar 7 Dept. of CSE

Chapter 3

Literature Survey And Motivation

[1] Various machine learning techniques, such as Decision Trees, Support Vector Machines,
and Neural Networks, were used in Huang’s 2019 study, "Machine Learning Algorithms for
Anticipating Learning Styles," to predict students’ learning styles based on VARK. The study
evaluated the efficiency of these algorithms in classifying learning styles using a dataset that
was acquired from a university. The results showed that the accuracy rates were encourag-
ing, and that Decision Trees performed better than other approaches.

[2] Singh et al. (2022) combined findings from a number of studies on machine learn-
ing algorithms for learning style prediction to create a thorough evaluation. The methods,
datasets, and effectiveness of several algorithms in predicting VARK learning styles were
summarized in the paper. The relevance of context-specific datasets and the importance of
feature selection were highlighted as ways to improve ability of individualized learning rec-
ommendations based on individual preferences.

[3] A. Masri and M. Al-Jabi, this paper have methodology is that these include capturing
the user’s image using a camera, segmenting the user’s body, overlaying virtual clothing

8
 Student Learning style prediction based on VARK using Machine Learning

items onto the user’s image, and providing real-time visualization of the virtual try-on ex-
perience. The authors implemented and evaluated their methodology, demonstrating the
feasibility and potential of the virtual dressing room application for enhancing the online
shopping experience. And have some advantages and disadvantage like Real-time try-on ex-
perience, User-friendly interface, Feasibility and practicality and Limited evaluation, Lack
of scalability consideration, Hardware and technical requirements, respectively.

[4] "Comparative Analysis of Machine Learning Algorithms for Predicting Learning Styles
Based on VARK Model," a study conducted by Kumar et al. (2021), assessed the ability of
several machine learning algorithms, such as Gradient Boosting, Random Forest, and Logis-
tic Regression, to predict VARK learning styles. Based on the analysis of learning behavior
data and preferences from a student sample, Random Forest was proven to perform excep-
tionally well in properly distinguishing learning styles.

[5] The topic of Gupta and Bhavsar’s 2022 study, "Predicting Students’ Learning Styles Us-
ing VARK and Machine Learning Techniques for Enhancing Personalized Learning," was
how to use learning style data based on VARK to predict students’ preferences. The research
classified students based on their learning methods using machine learning techniques like
K-Means Clustering and Neural Networks, opening the door to the possibility of individual-
ized learning recommendations based on individual preferences.

RIT, Rajaramnagar 9 Dept. of CSE

Chapter 4

Proposed System and Requirement

Specification

4.1 Proposed System

The suggested solution makes use of machine learning techniques to forecast student
learning patterns by utilizing the VARK model. First, input data is collected using a question-
naire that aims to capture important components of preferences related to visual, auditory,
reading/writing, and kinesthetic. After processing, the data is fed into a machine learning
model that has been trained on a variety of datasets to identify correlations and patterns
that represent various learning styles. The methodology allows for individualized learn-
ing recommendations by grouping pupils into distinct VARK categories. These suggestions
include individualized lesson plans, exercises, and evaluation styles based on the learning
preferences of each student. The system’s objective is to improve learning effectiveness and
engagement by giving teachers practical information to modify their approach to meet the
needs of each unique student.

10
 Student Learning style prediction based on VARK using Machine Learning

4.2 Software Requirements Specification-SRS

4.2.1 Modules

• Data Collection Module: Captures information relevant to the VARK learning styles
(Visual, Auditory, Reading/Writing, Kinesthetic) by administering questionnaires and
tests to students.

• Preprocessing Module: Makes sure the gathered data is ready for machine learning
by cleaning and preprocessing it. This covers feature extraction, normalization, and
handling of missing values.

• Machine Learning Module: Utilizing preprocessed data, algorithms are used to train
prediction models, finding correlations and patterns that support VARK learning styles.

• Prediction Module: Based on fresh input data, the trained model predicts the learning
preferences of specific students.

• Recommendation Module: Based on the anticipated learning styles, this module

offers individualized learning recommendations, including activities, materials, and
evaluation formats.

• Evaluation module: Allows for ongoing system enhancement by evaluating the pre-
cision and potency of forecasts and suggestions.

4.2.2 Functional Requirements

• Input Data Collection: The system has to let users enter a variety of information,
such as preferences, assessments, and behavioral patterns linked to learning styles.

RIT, Rajaramnagar 11 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

• VARK Model Integration: Integration of the Visual, Auditory, Reading/Writing, and

Kinesthetic (VARK) Model: This method of classifying learning styles involves inte-
grating the VARK model into the system.

• Machine learning algorithms: Application of different machine learning algorithms

for learning style prediction, such as neural networks, decision trees, and support vec-
tor machines.

• Prediction Accuracy: Make sure you accurately forecast the preferred learning styles
of your students by rigorously training and validating your system.

• Pattern Recognition: To improve forecast accuracy, the system should find correla-
tions and patterns in the incoming data.

• Classification of Learning Styles: Divide students into groups according to the learn-
ing style that they prefer (visual, auditory, reading/writing, or kinesthetic).

• User Interface: Provide a user-friendly interface that makes data entry, interaction,
and reading of prediction results easy.

• Performance Optimization: By refining algorithms and data processing methods,

you may guarantee that the system operates well even when dealing with big datasets.

• Security Measures: Put security measures in place to protect user information and
guarantee adherence to privacy laws.

4.2.3 Non-Functional Requirements

• Performance: In order to provide timely learning style predictions, the system must
be able to analyze big datasets with minimal reaction times, allowing for smooth user
interactions.

RIT, Rajaramnagar 12 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

• Scalability: In order to meet the rising demands of education, it must be able to handle
a growing number of users and data inputs without sacrificing forecast accuracy or
system performance.

• Reliability: Across a range of learning contexts and user demographics, the system
should produce consistent and predictable predictions, minimizing errors and guaran-
teeing dependable functionality.

• Security: Throughout the data handling and prediction processes, strong security mea-
sures must protect sensitive user data and uphold confidentiality, integrity, and avail-
ability.

• Usability: A user-friendly interface should be simple to use and intuitive, allowing

users with different levels of technical expertise to interact effectively.

• Accessibility: The virtual cloth trial room should be accessible to users with disabil-
ities. It should comply with accessibility standards, providing options for users with
visual impairments or other disabilities to interact with the system effectively.

• Compatibility: In order to promote widespread acceptance and smooth integration

into already-existing educational ecosystems, compatibility with a variety of devices
and platforms should be guaranteed.

4.3 Significance of the project

The Student learning style prediction model holds significant importance for students. Fol-
lowing are some key points highlighting its importance:

RIT, Rajaramnagar 13 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

1. Personalized Education: By providing individualized learning experiences catered

to each learner’s unique learning style, the initiative has the potential to significantly
transform education. It can correctly identify how pupils prefer to take information-
visually, auditorily, kinesthetically, or by reading and writing—by utilizing the VARK
model and machine learning algorithms.

2. Optimal Resource Allocation: Teachers can allocate resources more skillfully when
they are aware of the learning styles of their students. Resources like lesson plans,
exercises, and tests can be tailored to each student’s chosen learning style instead of
using a one-size-fits-all strategy, increasing comprehension and engagement.

3. Better Learning Outcomes: The initiative seeks to improve learning outcomes by

taking into account a variety of learning styles. When concepts are taught in the way
that they prefer to learn, students are more likely to retain them, understand them, and
succeed academically.

4. Improved Teacher-Student Relationships: The project helps teachers and students

understand each other better. By learning about each student’s unique learning prefer-
ences, teachers can modify their pedagogical approaches and build more relationships
and connections with their students in the classroom

5. Future Educational Innovations: If this initiative is implemented successfully, it

may open the door for more advancements in pedagogy and educational technology
in the future. It establishes a standard for utilizing psychological models and machine

RIT, Rajaramnagar 14 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

learning to enhance the teaching and learning process, which could completely trans-
form the educational landscape.

Applications:

1. Personalized Education: Teachers may tailor each student’s learning experience by

precisely anticipating their learning preferences through the use of machine learning
algorithms and the VARK model. This guarantees that instructional strategies cor-
respond with student preferences, resulting in increased comprehension and engage-
ment.

2. Adaptive Learning Platforms: The project’s results can be incorporated into these
platforms, which modify content delivery dynamically in response to anticipated learn-
ing methods. This enables a more effective and efficient learning process that meets
the various demands of the pupils.

3. Curriculum Design Optimization: By examining trends and connections in learning

style data, curriculum designers may make informed decisions. In order to promote
inclusive and productive learning environments, educators might modify lesson plans,
assignments, and evaluations to account for different learning styles.

4. Student Support Services: Learning style forecasts can direct the provision of fo-
cused help by student support services. For instance, additional visual materials or
aids may be provided to kids who have been classified as visual learners to help with
comprehension.

RIT, Rajaramnagar 15 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

5. Professional Development: Teachers can improve their teaching methods by using

the project’s findings. Comprehending the varied learning styles of their pupils allows
teachers to implement adaptable teaching strategies, which in turn enhances their ef-
fectiveness and the results they achieve for their students.

4.4 Scope of Project

The "Student Learning Style Prediction based on VARK Model using Machine Learning"
project’s scope includes a number of important elements. First, it entails utilizing the VARK
model (Visual, Auditory, Reading/Writing, Kinesthetic) to collect data on individuals’ pre-
ferred learning modes. The next step is using machine learning algorithms to examine this
data and create prediction models.
The goal of these models is to appropriately categorize pupils according to various learning
styles. To improve forecast accuracy, the research also seeks to find patterns and correlations
in the data. In order to maximize educational outcomes for specific pupils, it also entails
implementing tailored learning recommendations based on expected learning patterns.

4.5 Deployment Requirements

4.5.1 Hardware Requirements

For model deployment, a standard computing hardware setup is required. This in-
cludes a computer with sufficient processing power and memory capacity to handle
the computational demands of machine learning algorithms.

Minimum Hardware Specifications:

RIT, Rajaramnagar 16 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

1. A multi-core processor (e.g., Intel Core i5 or AMD Ryzen 5).

2. 8 GB or higher of RAM.
3. At least 250 GB of HDD storage space.

4.5.2 Software Requirements

• Python: The programming language used for coding and implementing machine learn-
ing models. Ensure Python is installed on the deployment environment.

• Integrated Development Environment (IDE): Jupyter Notebooks or any Python IDE

can be used for development and testing purposes.

• Required Libraries: Install necessary Python libraries such as Pandas, NumPy, and
Scikit-learn for data manipulation, analysis, and machine learning tasks

4.6 Others

An active internet connection is needed to access external dependencies such as data

sources, model updates, and software updates. Ensure reliable internet connectivity for un-
interrupted operation of the system.

4.7 Project Cost Estimate

The project cost estimate encompasses various expenses associated with hardware, soft-
ware, and potential data acquisition. Here’s a breakdown of the major cost components:

RIT, Rajaramnagar 17 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

1. Hardware Costs: Procurement of standard computing hardware, such as PCs or

servers, is essential for both development and deployment phases. Costs may vary
based on the spe- cific hardware requirements, including processing power, memory,
and storage capacity.

2. Data Acquisition Costs: Acquiring additional datasets or accessing premium data

sources for model training and testing purposes may involve expenses. These costs
could vary sig- nificantly depending on factors such as the complexity and volume of
the data required, as well as any licensing fees associated with proprietary datasets.

4.8 Project Deliverable

The "Student Learning Style Prediction Based on VARK Using Machine Learning" project
deliverables comprise a comprehensive dataset that includes learning style classifications
and student responses to VARK questionnaires. Based on input data, a trained machine
learning model that can reliably predict each learner’s preferred method (visual, auditory,
reading/writing, or kinesthetic) is created and verified.
Learning style factors are better understood because to the comprehensive reports on
patterns and correlations found in the training data. Additionally, the project produces an
intuitive dashboard or application that assigns students to the proper VARK categories and
provides individualized learning recommendations, including relevant readings, exercises,
and assessment types. Furthermore, easy replication and future research are ensured by
comprehensive documentation that covers the technique, algorithms employed, model per-
formance, and user advice.
Lastly, to encourage successful and individualized learning experiences, training ses-
sions and support materials are provided for educators to incorporate and reap the benefits
of the predictive model in their instructional design.

RIT, Rajaramnagar 18 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

4.9 Project Success

1. Accurate Learning Style Prediction: Based on the VARK model (Visual, Auditory,
Reading/Writing, Kinesthetic), machine learning techniques are used to accurately
predict the learning styles of individual pupils.

2. Pattern Recognition: Proficiently recognizing patterns and correlations in the training

data, improving the accuracy and dependability of learning style forecasts.

3. Effective Classification: Correct identification and categorization of each student’s

preferred learning style through efficient grouping of learners into VARK groups.

4. Personalized Learning Recommendations: Instructional materials, activities, and

assessment formats that are specifically designed to meet the individual learning pref-
erences of each student.

5. Enhanced Educational Experiences and Improved Learning results: Personalized

and adaptive teaching strategies that lead to better learning results.

6. Scalability and Adaptability: Creating a system that is adaptable to different learning

environments and learner populations.

7. Data-Driven Insights: Giving teachers practical information to improve their meth-

ods of instruction through the use of learner feedback and predictive analytics.

Participation in Project Competitions: Our project has achieved notable success through
active participation in various prestigious project competitions, underscoring its significance
and innovation. Here are the key competitions in which our project has been featured:

RIT, Rajaramnagar 19 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

‘Quantum 2k23’ - National Level Project Competition: ’Quantum 2k23’, a National

Level Project Competition sponsored by Bharat Forge, was scheduled on 7th December,
2023, and it marked a significant milestone. With great excitement, we took part in this
prestigious event and showcased our project to a diverse group of individuals, including
academics, industry professionals, and peers. Our goal was to be acknowledged for our
innovative approach to estimate proper learning style of Student using VARK model.

RIT, Rajaramnagar 20 Dept. of CSE

Chapter 5

Design

5.1 System Architecture

Figure 5.1: System Architecture of Learning style prediction model

21
 Student Learning style prediction based on VARK using Machine Learning

The figure 5.1 presents a methodical approach to learner style prediction. The first step
involves gathering information from learner activity logs and surveys, which are essential for
determining learning preferences. Preprocessing is the step where this raw data is cleaned,
changed, and arranged in preparation for analysis. After processing, the data is sent into
machine learning classifiers, which are taught to identify patterns that correspond to different
learning styles. After training, the accuracy of the model is verified, and if it is found to
be accurate enough, the procedure is terminated. But if more improvement is required, the
model is retrained with new data. By ensuring that the model is accurate in predicting learner
styles, this iterative process improves the efficacy and customisation of learning experiences.

RIT, Rajaramnagar 22 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

5.2 Data Flow Diagram

Figure 5.2: Data Flow diagram of Learning style prediction model

RIT, Rajaramnagar 23 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

The figure 5.2 depicts the process flow for predicting outcomes using machine learning
techniques:

1. Data Collection: Involves gathering raw data from various sources such as databases,
files, or sensors.

2. Data Cleaning: The collected data is cleaned to remove inconsistencies, errors, or

missing values.

3. Feature Extraction: Extracts relevant features from the dataset to use as input for the
machine learning model.

4. Feature Classification: Classifies the features based on their importance and rele-
vance to the target variable.

5. Split Data: Divides the dataset into training and testing datasets to train and evaluate
the machine learning model.

6. Prediction Model: Develops and trains the machine learning model using the training
dataset.

7. Machine Learning Process: Utilizes various machine learning algorithms to learn

patterns and relationships within the data.

8. Measure Accuracy: Evaluates the performance of the trained model by measuring its
accuracy on the testing dataset.

9. Accuracy Prediction: Predicts the accuracy of the model’s predictions based on the
evaluation results.

RIT, Rajaramnagar 24 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

This Data Flow Diagram illustrates the sequential flow of processes involved in predict-
ing outcomes using machine learning, from data collection to model evaluation and accuracy
prediction. Each step contributes to the development and refinement of the predictive model
to achieve accurate and reliable results.

5.3 UML Class Diagram

Figure 5.3: UML Class Diagram of Learning style prediction model

RIT, Rajaramnagar 25 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

The UML class diagram 5.3 for the "Student Learning Style Prediction Based on VARK
Model Using Machine Learning" project outlines the key classes and their interactions. The
Student class holds student details and their learning styles. The LearningStyle class repre-
sents individual learning styles with type and score attributes. The VARKSurvey class con-
ducts surveys using Question class instances to determine learning styles. The MLModel
class trains on student data to predict learning styles, while the DataHandlerclass manages
data loading and preprocessing. This structure ensures efficient prediction of students’ learn-
ing preferences by integrating survey data and machine learning.

RIT, Rajaramnagar 26 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

5.4 Use Case Diagram

Figure 5.4: Use Case Diagram

The use case diagram for the "Student Learning Style Prediction Based on VARK Model
Using Machine Learning" project encompasses key preprocessing and model-building stages.
Preprocessing involves preparing the data for analysis, ensuring it is clean and formatted cor-
rectly. Data Preparation follows, where data is organized and divided into training and testing
sets. Feature Projection and Feature Selection refine the dataset by identifying the most rel-
evant variables. Feature Scaling standardizes data ranges for consistency. Model Selection
entails choosing the most appropriate machine learning algorithm. Finally, the system makes
a Prediction based on the trained model, providing insights into students learning styles. This
systematic approach ensures accurate and meaningful predictions.

RIT, Rajaramnagar 27 Dept. of CSE

Chapter 6

Development/Implementation Details

6.1 Technology Used

6.1.1 Graphical user interface (GUI)

With the use of the "Student Learning Style Prediction" project’s graphical user interface
(GUI), students can enter data, see how their anticipated VARK learning styles look, and get
tailored learning suggestions. Teachers can analyze forecasts, gain insights, and offer input
to improve the efficacy and accuracy of the system.

6.1.2 Python

Python is commonly used to develop AI applications, such as improving human to com-

puter interactions, identifying trends, and making predictions. It offers easy-to-use and flex-
ible tools, is extensible, has a large number of libraries, and a vast community of Python
developers.

28
 Student Learning style prediction based on VARK using Machine Learning

6.1.3 OpenCV

A popular library for image processing and computer vision applications is called OpenCV
(Open-Source Computer Vision Library). It offers a variety of features for filtering, feature
extraction, and image manipulation—all crucial steps in the preprocessing of the taken pic-
tures.

6.1.4 Visual Studio Code

Microsoft created Visual Studio Code, or VS Code, a source-code editor for Windows,
Linux, and macOS. It is a simple yet effective source code editor that supports TypeScript,
JavaScript, and Node.js out of the box. It offers a wide range of extensions for many runtimes
and languages, including PHP, C++, Go, Java, and Python. Debugging, syntax highlighting,
intelligent code completion, snippets, code refactoring, and integrated Git are among the
features.

6.2 Data Processing and Analysis

6.2.1 Data collection

Data collection for the project involves gathering student learning data through surveys
and assessments. This data includes responses to VARK questionnaires, academic records,
and behavioral observations, which will be used to identify individual learning styles and
train the machine learning model for accurate predictions.

RIT, Rajaramnagar 29 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

Figure 6.1: Data collection sheet

6.2.2 Data Pre-processing

Raw data must be cleaned and transformed as part of the Student Learning Style Prediction
project’s data preprocessing. In order to make sure the dataset is appropriate for machine
learning methods, this involves resolving missing values, normalizing data, and encoding
categorical variables. This will ultimately improve model accuracy and performance for
predicting VARK learning styles.

Figure 6.2: Numeric Output after Data Preprocessing

RIT, Rajaramnagar 30 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

6.2.3 Data Splitting

In the project "Student Learning Style Prediction Based on VARK Model Using Machine
Learning," data splitting divides the collected data into training and testing sets. This ensures
the model is trained on one subset and validated on another, enhancing its accuracy and
robustness in predicting learning styles.

Figure 6.3: Splitting of Data into training and testing

6.3 Machine Learning Algorithms

6.3.1 Supervised Classification algorithms:

1. Logistic Regression

First, predictions are made with the logr.predict() method. Then, accuracy score com-
putes overall accuracy, while precision score and recall score calculate precision and recall
with a weighted average. The values are showing accuracy (0.5000), precision (0.5111), and
recall (0.5000). Additionally, a classification report provides detailed precision, recall, and
F1-score for each class, offering a comprehensive view of the model’s performance.

RIT, Rajaramnagar 31 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

Figure 6.4: Implementation of Logistic Regression

2. Decision tree

Figure 6.5: Implementation of Decision tree

RIT, Rajaramnagar 32 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

The project’s decision tree classifier. Importing the required libraries and initializing
the classifier are the first steps. The training data is then used to train the model. Recall,
accuracy, and precision are computed on test data using the trained model. The model’s
performance is shown by the accuracy, precision, and recall scores that are displayed in
the output. A classification report is also produced, which includes comprehensive stats for
every class. In this case, the model obtained 0.3636 accuracy, 0.4343 precision, and 0.3636
recall.

3. Random Forest

Figure 6.6: Implementation of Random Forest

A Random Forest classifier’s performance is assessed by the code through the use of
metrics such as accuracy, precision, and recall. The percentage of accurately predicted cases
is measured by accuracy. Recall quantifies the model’s capacity to recognize pertinent oc-
currences, whereas precision shows the accuracy of positive predictions. The classifier’s
accuracy in this case was 45.45, meaning that 45.45 of the predictions were accurate. Recall

RIT, Rajaramnagar 33 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

of 45.45 shows that the classifier was effective in identifying 45.45 of relevant occurrences,
while precision of 42.05 indicates that 42.05 of positive predictions were accurate.

4.KNN (K-Nearest Neighbors)

Figure 6.7: Implementation of KNN(k-Nearest Neighbors)

The accuracy of 0.3636 indicates that the k-Nearest Neighbors (KNN) classifier cor-
rectly predicted approximately 36.36 of instances in the test set. Precision of 0.3333 sug-
gests that around 33.33 of positive predictions made by the classifier were accurate. Recall
of 0.3636 implies that the classifier successfully identified approximately 36.36 of relevant
instances. These metrics collectively highlight the performance of the KNN classifier, pro-
viding insights into its effectiveness in correctly classifying instances in the dataset.

RIT, Rajaramnagar 34 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

5. Support Vector Machine (SVM)

Figure 6.8: Implementation of Support Vector Machine (SVM)

The output suggests that the Support Vector Machine (SVM) algorithm achieved an
accuracy of 36.36, meaning 36.36 of predictions were correct. The precision score of 33.33
indicates that only 33.33 of positive predictions were accurate, while the recall score of 36.36
suggests that the model successfully identified 36.36 of relevant instances. These metrics
collectively depict the SVM’s performance, highlighting its limited precision in correctly
identifying positive instances and its moderate ability to recall relevant instances within the
dataset.

RIT, Rajaramnagar 35 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

6. Naive Bayes

Figure 6.9: Implementation of Naive Bayes

The output given is in line with the performance indicators of a classifier that was
trained with the Naive Bayes technique. With an accuracy score of 0.3182, 31.82 of the
predictions came true. With a precision score of 0.4481, 44.81 of positive predictions appear
to have been correct. In addition, the recall score of 0.3182 indicates that 31.82 of pertinent
events were effectively detected by the model. The model’s overall accuracy and recall are
quite poor, even with a modest precision, suggesting that there may be opportunity for im-
provement in accurately forecasting instances and recognizing pertinent examples within the
dataset.

RIT, Rajaramnagar 36 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

6.3.2 Unsupervised Classification Algorithms:

1. K-Means Clustering

Figure 6.10: Implementation of K-Means Clustering

Performance of K-Means clustering:

• Silhouette Score (K-Means): 0.176070772440776

• Davies-Bouldin Index (K-Means): 1.3222783626978323

The performance metrics for the K-Means clustering algorithm indicate moderate clus-
tering quality. The Silhouette Score of 0.176 suggests that the clusters are somewhat sepa-
rated but overlapping, with values closer to 1 indicating better-defined clusters. The Davies-
Bouldin Index of 1.322 implies reasonable cluster compactness and separation, where lower

RIT, Rajaramnagar 37 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

values signify tighter, more distinct clusters. While the scores indicate acceptable clustering
performance, further optimization may be required to enhance cluster quality and achieve
more distinct cluster boundaries for improved data segmentation and analysis.

Graphical Representation of K-Means Clustering:

Figure 6.11: Clusters of Customer

Figure 6.11 indicates Graphical representation of K-Means clustering visually depicts

data points grouped into distinct clusters. Each cluster is represented by a centroid, with data
points assigned to the nearest centroid. Scatter plots or heatmaps illustrate cluster bound-
aries, aiding in understanding patterns and cluster separation for effective data analysis and
interpretation.

RIT, Rajaramnagar 38 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

2. Hierarchical Clustering

Performance of Hierarchical Clustering:

• Silhouette Score (KMeans): 1.315573031

• Davies-Bouldin Index (KMeans): 0.186455963

Graphical Representation of Hierarchical Clustering:

Figure 6.12: Hierarchical Clustering Dendrogram

Figure 6.12 represents dendrogram which is visual representation of hierarchical clustering

results. It displays the arrangement of data points in tree-like structure, illustrating the merg-
ing of clusters at each step of the clustering process. The height of the branches indicates the
distance or dissimilarity between clusters, aiding in cluster interpretation and analysis.

RIT, Rajaramnagar 39 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

Figure 6.13: Hierarchical Clustering result

Figure 6.13 represents Hierarchical clustering organizes data into a tree-like structure
where clusters at each level merge based on similarity. Results show hierarchical relation-
ships among data points, highlighting how clusters form progressively from individual ele-
ments to larger groups, aiding in understanding data structure and relationships effectively.

RIT, Rajaramnagar 40 Dept. of CSE

Chapter 7

Testing

7.1 Test Cases

Testing is the process of debugging a programme, which is one of the most important
components of computer programming. It is the process of evaluating a system or its com-
ponent(s) with the goal of determining whether or not it meets the defined requirements.
Testing can be done manually or automated, and it can include a variety of different types of
tests, such as unit tests, integration tests, and acceptance tests. It is putting in place a system
to discover any gaps, flaws, or missing requirements between what the country wants and
what it needs.The goal of testing is to identify any errors, bugs, or other issues in the system,
so that they can be corrected before the system is released to users. Features to be tested
in our application.In the project of predicting student learning styles based on the VARK
model using machine learning, test cases are crucial. They validate model accuracy across
diverse student profiles, ensuring robust predictions for visual, auditory, reading/writing, and
kinesthetic learning styles. Each test case covers varied data scenarios to verify model per-
formance comprehensively.

41
 Student Learning style prediction based on VARK using Machine Learning

Table 7.1: Student Learning style based on VARK model

Test Preconditions Steps Input Expected Actual Re- Status
ID Result sult
TC User Regis- 1.Navigate "John Doe", User is reg- User is reg- PASS
01 tration - User to the reg- Email: istered and istered and
is not logged istration "john@exa redirected redirected
in. page. 2. mple.com", to the login to the login
Fill in the Password: page. page.
registra- "pass-
tion form word123"
with name,
email, and
password.
3. Submit
the form.
TC Data Input 1. Navigate VARK ques- Data is suc- Data is suc- PASS
02 by Student to the data tionnaire re-cessfully cessfully
- User is input page. sponses. submitted submitted
logged in. 2. Click on and saved and saved.
Login but- in the sys-
ton. tem.
TC Data Pre- 1. Trigger Preprocessed Machine Machine PASS
03 processing the model training learning learning
- VARK training data. model is model is
question- module. trained suc- trained suc-
naire data is cessfully. cessfully.
submitted.
TC Model 1. Trigger Raw VARK Data is Data is PASS
04 Training the data question- cleaned and cleaned and
- Prepro- preprocess- naire data. formatted formatted
cessed data ing module. correctly correctly.
is available.
TC Learning 1. In- New VARK Learning Learning PASS
05 Style Pre- put new question- style is style is
diction - student naire re- predicted predicted
Machine data. 2. sponses. accurately. as "Visual".
learning Trigger the
model is prediction
trained. module.

RIT, Rajaramnagar 42 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

Table 7.2: Final Result

Test ID Preconditions Steps Expected Re- Actual Result Status
sult
TC 01 Learner Classifi- 1. Classify the Student is Student is PASS
cation - Learning student based classified into classified as
style prediction is on the predicted the correct "Visual".
completed. learning style. VARK cate-
gory.
TC 02 Personalized 1. Generate Appropriate Visual learn- PASS
Recommenda- personalized learning ma- ing materials
tions - Student is learning recom- terials and and activities
classified into a mendations. activities are are recom-
VARK category. recommended. mended.
TC 03 Feedback Col- 1.Collect feed- Feedback is Feedback is PASS
lection - Recom- back from the successfully successfully
mendations are student. 2. Sub- submitted and submitted and
provided to the mit feedback. saved. saved.
student.
TC 04 Valid input data 1.Input a large Timely com- Algorithm PASS
format dataset with pletion of completes
thousands of predictions predictions
student records without sys- within ac-
2. Execute the tem overload. ceptable time
prediction algo- frame.
rithm.
TC 05 Well-trained 1.Input learning Model identi- Prediction re- PASS
model with high data for a student fies dominant flects the pri-
accuracy. with mixed learn- learning style. mary learning
ing preferences. mode, consid-
2. Perform pre- ering the stu-
diction. dent’s diverse
preferences.
TC 06 Data preprocess- 1. Input clean Accurate pre- Prediction PASS
ing without miss- learning data. 2. diction of stu- matches the
ing values or out- Execute predic- dent’s learning student’s
liers. tion process. style. known learn-
ing prefer-
ences.

RIT, Rajaramnagar 43 Dept. of CSE

Chapter 8

Deployment

The "Student Learning Style Prediction" project is being deployed by creating a web-
based platform that both educators and students can access. The cloud service will host
the system in order to guarantee scalability and dependability. Utilizing machine learning
techniques, the backend will analyze incoming data, forecast learning preferences, and offer
tailored suggestions. To guarantee smooth updates and maintenance, pipelines for contin-
uous integration and deployment, or CI/CD, will be put in place. There will be security
measures in place to guarantee privacy and secure user data.

8.1 Readme File

A Readme file is a text file that contains information about a project, including instructions
for how to build and run the project. In the context of Virtual cloth trial room application,
a README file include information about how to install and configure the application as
well as any dependencies or prerequisites for the application. This file is included in the
root directory of a project, and is often one of the first things that a user will see when they
download or clone the project from a version control repository.

44
 Student Learning style prediction based on VARK using Machine Learning

8.2 User Manual

8.2.1 Home Page

1. Welcome screen with login/sign-up options

2. Overview of the VARK model

3. Navigation to key sections (Data Input, Results, Recommendations)

8.2.2 Recommendation Page

Displays personalized learning style prediction (Visual, Auditory, Reading/Writing, Kines-

thetic)

8.3 GUI for Analysis

The "Student Learning Style Prediction" project’s GUI analysis entails creating an in-
terface for students to enter their VARK (Visual, Auditory, Reading/Writing, Kinesthetic)
scores via quizzes or surveys. Using this information, the algorithm predicts the learning
style of the student. The graphical user interface (GUI) shows the prediction findings, em-
phasizing the prevalent learning style and offering customized suggestions. Educators can
also monitor prediction accuracy, examine student data, and provide comments to the ma-
chine learning model through the GUI. Upon submission, the GUI computes and displays
the final learning style prediction, ensuring clarity and usability. Effective GUI analysis
enhances user experience by providing clear insights into learning preferences, facilitating
personalized educational strategies tailored to individual students’ needs.

RIT, Rajaramnagar 45 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

RIT, Rajaramnagar 46 Dept. of CSE

Chapter 9

Results and Discussion

Based on the VARK model, the Random Forest algorithm predicted student learning
styles with the highest accuracy. Random Forest handled feature interactions and reduced
overfitting better than other algorithms, including SVM and KNN, as proven by extensive
testing on a variety of datasets. Because of its ensemble nature, which produced strong
and trustworthy predictions, it was perfect for use in educational settings. This better perfor-
mance highlights Random Forest’s ability to correctly detect individual learning preferences,
improving the tailored learning experience. Additional machine learning approaches and the
integration of more complex features could be explored in future research to further improve
the model.

This includes evaluation metrics such as accuracy, precision, and recall for each learning
style category. The discussion interprets these results, highlighting the model’s strengths
and limitations. Insights into how well the model predicts diverse learning preferences are
discussed, along with potential improvements or adjustments needed for better performance.
Overall, this section provides a comprehensive understanding of the project’s outcomes and
their implications for educational strategies tailored to individual learning styles.

47
 Student Learning style prediction based on VARK using Machine Learning

9.1 Classification Report of all algorithms

Figure 9.1: Result of Supervised Classification Algorithm

Figure 9.2: Result of Un-Supervised Classification Algorithm

RIT, Rajaramnagar 48 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

A variety of algorithms were examined to see how accurate they were at predicting
learning styles as part of the project "Student Learning Style Prediction Based on VARK
Model Using Machine Learning." The Random Forest algorithm showed the best accuracy
out of all of them. Random Forest was chosen for the GUI analysis phase of the project
because of its exceptional performance. This method is perfect for producing accurate pre-
dictions because of its robustness and versatility in handling different kinds of data. By
incorporating Random Forest into the graphical user interface, users are guaranteed to re-
ceive the most precise and reliable learning type predictions, which increases the overall
efficacy of the teaching aid.

Figure 9.3: Accuracy Comparison of Different Classifiers

RIT, Rajaramnagar 49 Dept. of CSE

Chapter 10

Conclusion and Future Work

10.1 Conclusion

It may be beneficial for students to understand their unique learning styles. This is cor-
roborated by the identification of the learning styles of the students and validated by several
studies that show the application of learning styles in conjunction with other learning strate-
gies improves academic performance or, at the very least, increases the enjoyment of study-
ing. The goal of this mixed-method study is to predict (with probability) the learning styles
of students who have mixed learning styles. In order to do this, ideas and techniques that
categorize students’ characteristics based on their preferred learning styles have been stud-
ied. The preferred learning styles were then predicted using a probabilistic method thanks
to the application of the Classification analysis. Here, the probability of learning styles was
predicted using regression using six machine learning techniques.

We chose a sample of one hundred and ten students to carry out our investigation. The
16-item VARK inventory questionnaire was used to gather the sample data, allowing four
distinct learning styles to be identified: visual (V), auditory (A), reading/writing (R), and

50
 Student Learning style prediction based on VARK using Machine Learning

kinesthetic (K). The best algorithm for forecasting the probability of each learning style was
the RF algorithm, according to the results. Using the same set of machine learning tech-
niques, we created classification models to predict the learning style label in order to test the
efficacy of our method. Our goal was to contrast our results with those of the classification-
based method. The results showed that all categorization models have comparatively low
accuracy. When predicting learning style, the RF had the highest accuracy, scoring 0.59

10.2 Future Work

1. Improved Model Precision:

• Increase the accuracy of your predictions by implementing more advanced ma-

chine learning techniques and algorithms.

• Make use of deep learning models to handle vast and complicated datasets more
effectively.

2. Expanded Data Sources:

• Incorporate extra data sources including engagement metrics, behavioral analyt-

ics, and real-time performance monitoring.

• Incorporate longitudinal data to document how learning styles evolve over time.

3. Enhancements to Personalization:

• Provide recommendation systems that are more intelligent and flexible, capable
of dynamically adapting to the input and progress of each unique learner.

RIT, Rajaramnagar 51 Dept. of CSE

 Student Learning style prediction based on VARK using Machine Learning

• Add elements of gamification to boost motivation and engagement.

4. Learning Management System (LMS) Integration:

• To make data collection and recommendation implementation more efficient, in-

tegrate the prediction system with well-known LMS platforms.

• Provide smooth data transfer between the prediction system and the current in-
structional tools.

5. Long-Term Research:

• To find out how personalized learning recommendations affect students’ perfor-

mance and satisfaction over time, conduct longitudinal research.

• Evaluate the degree to which academic success and retention rates correspond
with projected learning styles.

6. Validation Across Cultures:

• To guarantee the model’s global adaptability and efficacy, validate it in various

cultural and educational contexts.

• Make learning recommendations that are relevant and sensitive to cultural differ-
ences.

RIT, Rajaramnagar 52 Dept. of CSE

Chapter 11

References

[1] F. Rasheed and A. Wahid, “Learning style detection in E-learning systems using machine
learning techniques,” Expert Syst. Appl., vol. 174, 2021.
https://www.sciencedirect.com/science/article/abs/pii/S0957417421002153
[2] Varsha.T.Lokare and Prakash M Jadhav an AI-based learning style prediction model for
personalized and effective learning,(2023).
https://shorturl.at/cFP0T3
[3] Tang, J., Chaw, L. Y. (2019). Investigating learning styles, teaching styles and use of
technology for personalized learning in higher education. Education and Information Tech-
nologies, 24(5), 3243-3262.
https://shorturl.at/fLH2l
[4] Joyce, C., Hassall, T., Hassall, R. (2018). The VARK learning preferences inventory as
a tool for predicting academic achievement in psychology students. Psychology Teaching
Review, 24(1), 20-26.
[5] J. Bernard, T. W. Chang, E. Popescu, and S. Graf, “Learning style Identifier: Improving
the precision of learning style identification through computational intelligence algorithms,”
Expert Syst. Appl., vol. 75, pp. 94–108, 2017.

53
 Student Learning style prediction based on VARK using Machine Learning

https://shorturl.at/YClsX
[6] I. Azzi, A. Jeghal, A. Radouane, A. Yahyaouy, and H. Tairi, “A robust classification to
predict learning styles in adaptive E-learning systems,” Educ. Inf. Technol., vol. 25, no. 1,
pp. 437–448, Jan. 2020.
https://shorturl.at/t0ILU
[7] C. Romero and S. Ventura, “Educational data mining and learning analytics: An updated
survey,” Wiley Interdiscip. Rev. Data Min. Knowl. Discov., vol. 10, no. 3, p. e1355, May
2020.
https://shorturl.at/X9jjD
[8] Cunningham, P., Delany, S. J. (2021). K-nearest neighbour classifiers-a tutorial. ACM
Computing Surveys (CSUR), 54(6), 1-25.
[9] Nalepa, J., Kawulok, M. (2019). Selecting training sets for support vector machines: a
review. Artificial Intelligence Review, 52(2), 857-900.
https://link.springer.com/article/10.1007/s10462-017-9611-1
[10] Resende, P. A. A., Drummond, A. C. (2018). A survey of random forest based methods
for intrusion detection systems.
[11] Ashfaque, J. M., Iqbal, A. (2019). Introduction to support vector machines and kernel
methods.
[12] Zhang, S., Li, X., Zong, M., Zhu, X., Cheng, D. (2017). Learning k for knn classifica-
tion. ACM Transactions on Intelligent Systems and Technology (TIST).
[13] J. Xu, K. H. Moon, and M. Van Der Schaar, “A Machine Learning Approach for Track-
ing and Predicting Student Performance in Degree Programs,” IEEE J. Sel. Top. Signal
Process., vol. 11, no. 5, pp. 742–753, 2017.
[14] Towards an online continuous adaptation mechanism (OCAM) for enhanced engage-
ment: An EEG study International Journal of Human–Computer Interaction (2019).
https://rb.gy/cq938w

RIT, Rajaramnagar 54 Dept. of CSE

[15] Advanced prediction of learnerś profile based on Felder-Silverman learning styles using
web usage mining approach and fuzzy c-means algorithm International Journal of Computer
Aided Engineering and Technology (2019 ).
https://rb.gy/cq938w
[16] F. Muñoz-Bullón, M. J. Sanchez-Bueno, and A. Vos-Saz, “The influence of sports par-
ticipation on academic performance among students in higher education,” Sport Manag.
Rev., 2017.
[17] J. Xu, K. H. Moon, and M. Van Der Schaar, “A Machine Learning Approach for Track-
ing and Predicting Student Performance in Degree Programs,” IEEE J. Sel. Top. Signal
Process., vol. 11, no. 5, pp. 742–753, 2017.
[18] M. Shahiri, W. Husain, and N. A. Rashid, “A Review on Predicting Student’s Perfor-
mance Using Data Mining Techniques,” in Procedia Computer Science, 2015.[4] Y. Meier,
J. Xu, O. Atan, and M. Van Der Schaar, “Predicting grades,” IEEE Trans. Signal Process.,
vol. 64, no. 4, pp. 959–972, 2016.
[19] K. P. Shaleena and S. Paul, “Data mining techniques for predicting student perfor-
mance,” in ICETECH 2015 - 2015 IEEE International Conference on Engineering and Tech-
nology
[20] J. D. Kelleher, B. Mac Namee, and A. D’Arcy, Fundamentals of Machine Learning
for Predictive Data Analytics. Algorithms, Worked Examples, and Case Studies. 2015.
https://rb.gy/kunb0v

55
Chapter 12

Plagiarism Report

56
Chapter 13

Activity Chart

Figure 13.1: Activity Chart

57
Chapter 14

QUANTUM 2k23 Poster

Figure 14.1: Poster

58