Final Report 1

VISVESVARAYA TECHNOLOGICAL UNIVERSITY
“JNANA SANGAMA”, BELAGAVI, KARNATAKA-590018
A Project Report on
“ COURSE RECOMMENDATION AND SKILL PREDICTION

BASED ON CAREER CHOICE ”
By
ADARSH S G 4MH19IS002
AKSHATHA M 4MH20IS009
DHANUSH G R 4MH20IS027
RAKESH J 4MH20IS066
2023-2024
DEPARTMENT OF INFORMATION SCIENCE AND ENGINEERING
MAHARAJA INSTITUTE OF TECHNOLOGY, MYSORE
BELAWADI, NAGUVANAHALLY POST, S.R. PATNA TALUK,

MANDYA-571 477
DEPARTMENT OF INFORMATION SCIENCE & ENGINEERING
MAHARAJA INSTITUTE OF TECHNOLOGY MYSORE
MANDYA-571477
CERTIFICATE
Certified that the project work titled “COURSE RECOMMENDATION AND SKILL
PREDICTION BASED ON CAREER CHOICE” has been successfully carried out by
ADARSH S G [4MH19IS002], AKSHATHA M [4MH20IS009], DHANUSH G
R[4MH20IS027], RAKESH J [4MH20IS066] bonafide students of Maharaja Institute of
Technology, Mysore in partial fulfilment of requirements of Degree of Bachelor of
Engineering in Information Science & Engineering of Visvesvaraya Technological
University, Belgaum during the academic year 2019-2020. It is certified that all
corrections/suggestions indicated for the internal assessment have been incorporated in the
report deposited in the department library. The project report has been approved as it satisfies
the academic requirements with respect to the project work prescribed for Bachelor of
Engineering Degree.
Signature of guide Signature of the Principal Signature of the

HOD
Prof.Amruth V Dr.B.G.Naresh Kumar Dr.Sharath Kumar
Y.H
Assistant Professor Principal
Professor
Dept.of IS&E, MIT, Mysore &HOD of IS&E,
MIT, Mysore MIT, Mysore
Name of the Examiners External Viva Signature with date
1.
2.
ACKNOWLEDGEMENT
We sincerely owe our gratitude to all the persons who helped and guided us in completing this
technical seminar work
We are thankful to Dr. B.G. Naresh Kumar, Principal, Maharaja Institute of Technology
Mysore, for having supported us in our academic endeavors.
We are extremely thankful to Dr. Sharath Kumar Y H, Professor & Head, Department of
Information Science and Engineering, for his valuable support and his timely inquiries into
the progress of the work.
We are greatly indebted to our guide Prof.Amruth v, Asst. Professor, Department of

Information Science and Engineering, for the consistent co-operation and support.
We are obliged to all teaching and non-teaching staff members of Department of

Information Science and Engineering, for the valuable information provided by them in their
respective field’s. We are grateful for their co-operation during the period of our technical
seminar work.
ADARSH S G 4MH19IS002
AKSHATHA M 4MH20IS009
DHANUSH G R 4MH20IS027
RAKESH J 4MH20IS066
ABSTRACT
In today's digital age, online learning has emerged as a popular and convenient way
for individuals to acquire new skills, expand their knowledge, and pursue educational
opportunities from the comfort of their homes. With the proliferation of online course
platforms offering a vast array of courses across various domains, the challenge for
learners lies in navigating this abundance of options to find the most relevant and
suitable courses that align with their interests, goals, and learning preferences. This is
where the role of an Online Course Recommendation System becomes invaluable.
The primary objective of an Online Course Recommendation System is to enhance the

learning experience for users by providing them with personalized recommendations
that match their unique learning needs and preferences. By facilitating the discovery of
relevant and high-quality courses, these systems enable learners to make informed
decisions and optimize their learning journey. Furthermore, Online Course
Recommendation Systems contribute to increasing user engagement, retention, and
satisfaction on online learning platforms, ultimately fostering a positive and rewarding
learning experience for learners.
.
CONTENTS
Chapter 1 : INTRODUCTION Page-no

Overview 1
Problem Statement 1
The Solution 2
Existing System 2
Proposed System 3
Chapter 2 : LITERATURE SURVEY

Survey Papers 4
Survey Findings 17
Chapter 3 : SYSTEM REQUIREMENT SPECIFICATION

Stakeholders
Functional Requirements 23
Non-Functional Requirements 23
System Requirements 25
26
Chapter 4 : SYSTEM ANALYSIS AND DESIGN

System Analysis 27
High Level Design 28
Low Level Design 30
Chapter 5 : IMPLEMENTATION
Methodology and Techniques 31
Control Flow 36
Chapter 6 :TESTING
Design of Test Cases 37
Unit Testing 37
Integration Testing 38
Acceptance Testing 38
System Testing 39
Test strategy and Approach 39
Test Cases 40
SNAPSHOTS AND RESULT DISCUSSION 43

FUTURE ENHANCEMENT 47
CONCLUSION 48
REFERENCES 49
vi
Course Recommendation and Skill Prediction Based on Career Choice 2023-2024
Chapter 1
INTRODUCTION
Overview
The Course Recommendation and Skill Prediction system (CRSP) is a
sophisticated tool designed to guide individuals in making informed decisions about their
education based on their chosen career paths. Using advanced machine learning
algorithms, CRSP assesses user-provided career aspirations, industry trends, and job
market data to generate personalized course recommendations. By analyzing historical
job postings, CRSP predicts future skill demands, allowing users to identify areas for skill
enhancement. The system seamlessly integrates a vast database of courses from both
traditional and online platforms, employing a refined matching algorithm to provide a
ranked list of recommended courses. CRSP also emphasizes user feedback for continuous
improvement and a dynamic learning experience. Privacy and data security are
paramount, with stringent measures in place to safeguard user information. CRSP
empowers individuals to proactively shape their professional development, ensuring that
their educational pursuits are aligned with the demands of their chosen career path, thus
enabling lifelong learning and sustained career growth.
Problem statement
Every organization wants to be success. Proper plan and suitable human resources or
capable human resources may lead the organization success. It is very important to know
the right candidate for the right job and also factors associated with job types. There is no
automation for identification of factors associated with jobs in real time which may lead
to low candidate performance and loss to the organization. A proper automated system
required which proposed system does using machine learning algorithms.
Dept of IS&E, MITM 1

The solution
Course Recommendation and Skill Prediction (CRSP) employs a multi-faceted
approach. Natural Language Processing (NLP) is used to extract and categorize user-
provided career preferences. Machine learning algorithms analyze industry trends, job
market data, and historical job postings to predict future skill demands. A matching
algorithm considers course content, duration, and user feedback in the extensive course
database. Adaptive learning techniques refine recommendations based on user
interactions. Data encryption and strict access controls ensure privacy and security.
Continuous integration with reputable data sources keeps recommendations up-to-date.
This comprehensive framework leverages advanced technologies to empower individuals
in making informed decisions about their educational pursuits aligned with their chosen
career .
Existing System
Student management system – tool which maintains student academic details such as
marks, attendance, admission, fees, sports, placement cell etc. but doesn’t provides any
useful information related to recruitment process.
General Counseling – this is manual system where college staffs will sit and conduct
counseling for the students to know the issues they are facing. This is manual process
and too time consuming.
Human Ad-visors – this system is also manual where experts gives their opinions,
suggestions for the issues. Manual process which requires time, experience and money.
Recruitment process and also identifying jobs factors which suits job type is a complex
process in any organization. Current recruitment process requires more time, more
money, human resources and advertisement etc. In the current system there is no system
to identify factors for the jobs and also online course recommendation. Complete
recruitment process is manual. It is a very challenging task for HR team to identify the
suitable applicants for the jobs. Few existing research papers just developed machine
learning models and shown results of around 90%, but these works are not suitable for
real time. We require a system that can automatically identifies factors associated with

job types and also online course recommendation for the final year students.
Limitations of Existing System: It is clear from the details of existing system that the
whole system is complex and required a lot of manual work.
 Only Model built.
 No real time implementations done.
 Manual process
 More time and money required
 Human resources and advertisements required
Proposed System
 System has 2 major objectives online course recommendation for the students and
predicting the relationship between educational factors with job types.
 System developed as real time application meant for the companies.
 System is a browser based application where end users can use browsers such as
chrome, edge, and opera to access the project.
 System uses both supervised learning algorithms and unsupervised learning

algorithms to process datasets and for prediction.

Chapter 2
LITERATURE SURVEY
Survey Papers
The following papers/thesis/journals are referred for information regarding our

application.
1. Skill-based Career Path Modeling and Recommendation:
In this paper, the author proposes Skill-based Career Path Modeling and
Recommendation The development of new technologies at an unprecedented rate is
rapidly changing the landscape of the labor market. Therefore, for workers who want to
build a successful career, acquiring new skills required by new jobs through lifelong
learning is crucial. In this paper, we propose a novel and interpretable monotonic
nonlinear state-space model to analyze online user professional profiles and provide
actionable feedback and recommendations to users on how they can reach their career
goals. Specifically, we use a series of binary-valued and nondecreasing latent states to
represent the expanding skill set of each user throughout their career and propose an
efficient inference method under our model. Using a series of experiments on two large
real-world datasets, we show that our model (sometimes significantly) outperforms
existing methods on the tasks of company, job title, and skill prediction. More
importantly, our model is interpretable and can be used for other important tasks
including skill gap identification and career path planning. Using a series of case studies,
we show that our model can provide i) actionable feedback to users and guide them
through their upskilling and reskilling processes and ii) recommendations of feasible
paths for users to reach their career goals.
Disadvantages: The paper does not provide any user feedback or case study to
demonstrate the effectiveness or usability of the proposed system .
2. Online Course Recommendation Using Deep Convolutional Neural Network with

Negative Sequence Mining ;
In this paper the authors Ming Gao, Yonghan Luo, and Xiaonan Hu propose the

shortcomings of the high truancy rate and poor personalized recommendation effect in
traditional online learning platforms, this paper combines a convolutional neural
network with negative sequence pattern mining into online course recommendation and
proposes a personalized online course recommendation model based on convolutional
neural network combined with negative sequence pattern mining. The model can not
only recommend a list of courses with maximum demand shortly but also predict which
courses are most likely to be mis-selected by users. Experimental results show that the
recommendation model proposed in this paper has a certain degree of course
recommendation accuracy, and the method has a significant performance in predicting
miss-selected courses, which provides a new thought and insight for online course
recommendation.
Disadvantages: The paper does not provide any user feedback or case study to
demonstrate the effectiveness or usability of the proposed system.
3. A Job Recommendation Method Optimized by Position Descriptions and

Resume Information:
In this paper, the author Peng Yil, Cheng Yang, Chen Lil, and Yingya Zhangl
proposes the development of Internet technology, and online job-hunting plays an
increasingly important role in job searching. It is difficult for job hunters to solely rely
on keyword retrieving to find positions that meet their needs. To solve this issue, we
adopted an item-based collaborative filtering algorithm for job recommendations. In this
paper, we optimized the algorithm by combining position descriptions and resume
information. Specifically, the job preference prediction formula is optimized by
historical delivery weight calculated by position descriptions and similar user weight
calculated by resume information. The experiments tested on real data set have shown
that our methods have a significant improvement on job recommendation results.
Disadvantages: The accuracy of the predictions made using the data mining techniques
may be affected by the quality and completeness of the data used in the analysis

4. An Intelligent Career Guidance System using Machine Learning:

In this paper, the author Vignesh S, Shivani Priyanka C, Shree Manju H, and Mythili K
proposes Most of students across the world are always confused after they complete
higher secondary and the stage where they have to choose an appropriate career path. At
the age of, the students don't have adequate maturity to accurately know what an
individual has to follow in order to choose a congenial career path. As we pass through
the stages, we realize that every student undergoes a series of doubts or thought
processes on what to pursue after 12th which is the single tallest question. Then comes
the next agony of whether they have the essential skills for the stream they’ve chosen.
Our computerized career counseling system is used to predict the suitable department
for an individual based on their skills assessed by an objective test. If one completes
their online assessment which we have created in our system, then automatically they
will end up choosing an appropriate course which will also reduce the failure rate by
choosing a wrong career path.
Disadvantages: The system uses only two parameters. The system is applicable only to
B.Tech courses limited number of outcomes such as dream company, core company,
and mass recruiters, are not eligible.
5. Exploring Social Media Data for MOOC Recommendation:

In this paper, the author Sara Assam, Najima Daoudi, and Rachida Ajhoun proposes
MOOCs (Massive Open Online Courses) are a key tool for open education stakeholders
and high-level universities to face the problem of limited resources and make education
accessible worldwide. Our study relies on the hypothesis of the influence of learning
adaptation on the learners' lack of motivation in MOOCs. In this sense, a
recommendation approach could stimulate the learner's interest in specific MOOCs. In
previous research, we have identified the recommendation criteria to recommend
MOOCs adapted to the learner's needs and motivations. Then, we modeled the ontology
of the learning actors’ profiles for the matching of the characteristics of learners with
adapted MOOCs. Afterward, we designed the functional architecture and use cases of a
semantic recommendation system to stimulate learners' interest in MOOCs. However, the
recommendation process is still hindered by the learner profile on these platforms which
doesn’t represent the learner’s interests and motivations. Yet, social media profile

extraction can be an external data source for learners' profile enrichment. So, the purpose
of this research is to use social media mining for the acquisition of personal and
professional data about learners. Therefore, we apply the first phases of data mining on a
social media dataset to explore the potentially meaningful data for MOOC
recommendation.
Disadvantages: The proposed approach has technical limitations, requiring advanced text
mining techniques to address challenges in Natural Language Processing for social media
data
6. Educational Data Analytics using Association Rule Mining and Classification:
In this paper, the author Pornthep Rojanavasu proposes The education crisis is now
widely spread globally in terms of decreasing the number of students and decreasing
degree requirements for some jobs. Educational data mining (EDM) is recently interested
in the data mining area to discover useful knowledge in educational data to help educators
improve their administration planning and student services. This paper proposes applying
two data mining techniques to educational data. First, the association rule was applied to
admission data to find some knowledge for supporting admission planning. Second, a
decision tree was applied to course grades and job data of graduated students to predict
jobs after graduation. The results of these studies give good knowledge for admission
planning and job prediction.
Disadvantages: Limitation is related to the quality and quantity of the available data. Data
quality issues such as missing values or noise in the data can affect the accuracy of the
results generated by the system.
7. Engineering Students’ Job Performance Prediction Model based on Adversity Quotient

and career Interest:
In this paper the author M. F. Lee, N.F. Mat Nawi & C.S. Lai propose Job
performance of an individual can be predicted through personal highlights. The
capability of predicting job performance is benefiting from all perspectives. Job

performance in this study refers to the abilities of students in completing the task given
by the lecturer. Through this study, student’s job performance through Adversity
Quotient (AQ) and career interest were determined. A total of 370 Bachelor’s students
from eight faculties in UniversitiTun Hussein Onn Malaysia (UTHM) have been selected
as respondents by the sampling strata. The research was a survey study and the
instruments were a questionnaire of AQ Related Personality Traits (ARP) Stoltz,
Inventory Self-Directed Search Form Easy (SDSS) Holland, and Job Performance
Questionnaire. Data were analyzed using sum, mean, frequency, percentage, and
Multiple Linear Regression tests. The findings show that the majority of respondents
have in high level of AQ, and the dominant career interest of students is Social
personality. The Multiple Linear Regression analysis shows job performance is
predictable through the variables Adversity Quotient (AQ) and two types of personality
career interest (Social and Entrepreneur), that is Job Performance = 35.21 + 7.28 (S) +
3.98 (E) + 0.11 (AQ). This model is expected to be applied to Bachelor’s students of
UTHM to predict future job performance based on scores of AQ, Social, and
Entrepreneur .
Disadvantages: The proposed approach has technical limitations, requiring advanced text
mining techniques to address challenges in Natural Language Processing for social media
data.
8. Supporting Users of Open Online Courses with Recommendations: an

Algorithmic Study:
In this paper the author Soude Fazeli, and Enayat Rajabi proposes Almost all studies
on course recommenders in online platforms target closed online platforms that belong to
a University or other provider. Recently, a demand has developed that targets open
platforms. Such platforms lack rich user profiles with content metadata. Instead, they log
user interactions. We report on how user interactions and activities tracked in open online
learning platforms may generate recommendations. We use data from the Open U online
learning platform in use by the Open University of the Netherlands to investigate the
application of several state-of-the-art recommender algorithms, including a graph-based
recommender approach. It appears that user-based and memory-based methods perform

better than model-based and factorization methods. Particularly, the graph-based
recommender system outperforms the classical approaches on prediction accuracy of
recommendations in terms of recall.
Disadvantages: The paper acknowledges that the data istoo sparse and noisy to make
reliable recommendations with traditional approaches, and suggests using hybrid methods
or incorporating content information to improve the results .
9. HYBRID MASSIVE OPEN ONLINE COURSE (MOOC)

RECOMMENDATION SYSTEM USING MACHINE LEARNING:
In this paper, the author Vishal Garg, Dr. Ritu Tiwari proposes The
amount of educational resources spread over the Internet is huge and diverse. Massive
Online Open Courses(MOOC) websites such as Coursera, Udacity, and EdX are gaining
momentum. In this knowledge exploration age, it becomes increasingly difficult for a
learner to identify a suitable course for him. This task can be tedious because it involves
accessing each platform, searching available courses, selecting some courses, read
carefully each course syllabus, and choosing appropriate content. To counter this
problem Recommendation Systems are used. This paper proposes an effective and
efficient MOOC Recommendation System based on Machine Learning. The proposed
system will be able to provide effective recommendations to learners on e-learning sites.
Disadvantages: It does not evaluate the effectiveness or accuracy of the proposed
system on real-world data or users.
10. Course Recommendation of MOOC with Big Data Support A Contextual Online Learning
Approach:
In this paper the authors Yifan Hou, Pan Zhou, and Jie Xu propose With the advent
of the big data era of MOOC, enrolled students and offered courses become numerous
and diverse, resulting in a large amount of data and complex curriculum relationships.
Thus how to recommend appropriate coursesto improve students’ learning outcomes has
become a daunting task. The state-of-the-art works ignore some significant features in
course recommendation of MOOC: heterogeneity of large-scale user groups, sequence

problems in courses, and the foreseeable quantitative explosion of courses and users.
This paper proposes a systematic methodology for recommending personalized courses
while considering the sequence of learning curriculum. The system works by
recommending the course with the highest reward to a user. New feedback from the
user is then recorded and will be used to improve the performance of recommendations
for future students. The core component is a novel online learning algorithm based on
hierarchical bandits with known smoothness. We analyze the performance of our
proposed online learning algorithm in terms of regret and prove the asymptotic
optimality of the proposed algorithm
Disadvantages: It does not evaluate the effectiveness or accuracy of the proposed
system on real-world data or users.
11. Smart Career Guidance and Recommendation System:
In this paper the author Lakshmi Prasanna, DR.D.Haritha proposes Recommender

system is a computer program built with the help of experts where the details of the
students and their aptitudes help them find the right course for their future. This project
proposes feasible predictions for student’s field selection based on their marks and choice
of interest. Choosing the right field in the CSE/IT stream is very important for his/her
future. If the decision goes wrong it will be a mismatch between student aptitude,
capability, and personal interest. This project also reveals the research process for the
preparation of such a recommender system. Smart Career Guidance Recommendation
System is developed for recommending skilling courses and certification courses in the
CSE/IT domain. A substantial amount of literature focuses on predicting student
performance in solving problems or completing courses. Many Machine learning
techniques, such as decision trees artificial neural networks, matrix factorization,
collaborative filters, and probabilistic graphical models, have been applied to develop
student performance prediction algorithms. In this paper, we identify and apply suitable
algorithms for Student-specific skill-oriented course recommendation systems in the
CSE/IT domain.
Disadvantages: The paper does not compare the naive Bayes algorithm with other

existing algorithms or methods for career guidance
12. A Study of the Mass Customization-based Strategy for the Recommendation of

Online Course Resources of the Open University of China:
In this paper, the author YANG Jing proposes This article analyzes the contradictions
between teaching personalization and teaching standardization. By drawing upon the
concept of “mass customization” in the manufacturing industry, this article puts forward
the model of the mass customization of students’ personalized course systems based on
online courses. Mass customization is based on the modularization and standardization of
course resources, and the personalized course system is customized through the assembly
of course resource modules; learning resource modules are recommended to students in
the collaborative filtering mode, and the resource recommendation strategy is developed
based on modularization to help students quickly identify the course resources they are
interested in and complete the customization process.
Disadvantages: The system is designed to recommend courses based on similarities

between students and meets the specified minimum acceptable grade for the
recommended course
13. A Collaborative Recommendation system for online course recommendations:

In this paper, the author Raghad Obeidat, and Rehab Duwairi Ahmad Al-Aiad proposes
a collaborative recommender system that recommends online courses for students based
on similarities of students' course history. The proposed system employs data mining
techniques to discover patterns between courses. Consequently, we have noticed that
clustering students into similar groups based on their respective course selections plays a
vital role in generating association rules of high quality when compared with the
association rules generated using the whole set of courses and students. In particular, the
Apriori algorithm was used to generate association rules; once using the whole dataset
and once using the clusters which are formed based on students' choices of courses. The
results reveal that the coverage of the rules generated on clusters is better. Also, to assess
the effect of course dependency on recommendations, we applied the SPADE algorithm
to course sequences. The results are in harmony with the results obtained when Apriori

was applied.
Disadvantages: The system is designed to recommend courses based on similarities
between students and meets the specified minimum acceptable grade for the
recommended course
14. Predicting the Suitability of IT Students’ Skills for the Recruitment in Saudi
Labor Market:
In this paper the author Raghad Munirah Alghamlas and Reham Alabduljabbar propose
In a technical world, knowledge of information technology (IT) skills is an important
aspect, especially in Saudi Arabia, with the development in the technical labor market.
This paper proposed a solution to examining the demand for recruitment in the Saudi
market, using data extracted from online job portals. By utilizing machine learning
algorithms to develop a web-based application that helps in predicting the suitability of
IT students’skillsfor the recruitment in Saudi labor market. In addition, this solution
offers general recommendations of the highest needed IT skills in Saudi Arabia in order
to overcome the unemployment problem.
Disadvantages: The paper does not specify which machine learning algorithms are used
or how they are evaluated. The paper does not address the ethical or social implications of
using data mining techniques for career guidance.
15. Research on Optimal Design of Online Education Course Recommendation

System Based on Hybrid Recommendation Algorithm:
In this paper the author Yuxin Ren. Zhifang proposes In recent years, online
education technology has developed rapidly, and the market scale of online education
platforms is growing. However, with the widespread application and development of this
technology, the requirements of online education users are gradually becoming stricter. A
personalized course recommendation system can greatly improve the satisfaction and
learning efficiency of student users, making the education platform stand out from many
competitors. Based on the analysis of the advantages and disadvantages of the existing
learning resource recommendation system on the online education platform, this paper
proposes a new learning resource recommendation model based on a hybrid
recommendation algorithm which includes a course recommendation submodule based
on statistics and personalized course recommendation submodule based on professional
training requirements. It can not only help users to find high-quality information that they
are interested in, save users' time and cost, but also effectively solve the problem of cold
boot. This paper inspires the improvement of the online education course
recommendation system.
or how they are evaluated. The paper does not address the ethical or social implications
of using data mining techniques for career guidance
16. Massive open online course recommendation system based on a reinforcement

learning algorithm:
In this paper, the author Jian-Wei Tzeng, Nen-Fu Huang, An-Chi Chuang, Ting-
Wei Huang, and Hong-Yi Chang proposes Massive open online courses (MOOCs) are
open online courses designed based on teaching progress. Videos and learning exercises
are used as learning materials in these courses, which are open to numerous users.
However, determining the prerequisite knowledge and learning progress of learners is
difficult. Based on learners’ online learning trajectory, we designed a set of practice
questions for a recommendation system for MOOCs, provided suitable practice questions
to students through the LINE chatbot (a type of social media software), and used mobile
devices to encourage participation in MOOCs. Reinforcement learning, which involves
reward function design and iterative solution improvement, was used to set task goals,
including those related to course learning and practice question difficulty. The proposed
system encouraged certain learning behaviors among students. Students who used the
system exhibited an exercise completion rate of 89.97%, which was higher than that of
students who did not use the system (47.23%). The system also increased the student's
overall learning effectiveness. Students who used and did not use the proposed system
exhibited average midterm scores of 64.73 and 58.21, respectively. We also collected 227
online questionnaires from students. The results of the questionnaires indicated that 90%
of the students were satisfied with the system and hoped to continue using it.
Disadvantages: The paper uses a small and limited dataset of 100 students and 20

courses from a single university, which may not be generalizable to other contexts and
domains. The proposed approach does not include natural language processing.
17. Massive Open Online Courses (MOOCs) Recommendation Modeling using Deep
Learning:
In this paper, the author Siriporn Sakboonyarat, Panjai Tantatsanawong proposes

That knowledge in the world of the internet has always been developed with updated
information. Recommendation system for Massive Open Online Courses (MOOCs) can
help create endless learning opportunities. This study presents a Massive Open Online
Courses Recommendation Modeling using Deep Learning with Multilayer Perceptron
architecture which is suitable for enormous data analysis. The research methodology
begins with the process used for the data analysis process, using the data mining
technique according to the Cross-industry standard process for data mining (CRISP-DM),
consisting of six steps: business understanding, understanding of data, data preparation,
modeling, evaluation, and deployment. We received a set of data from Harvard and MIT,
published for edX learning data in 2012-2013, consisting of 16 programs, 18 features, and
641138 sample items. The research found that the most appropriate model is a model
with 7 hidden layers and a 1e-3 learning rate, processed by GPU acceleration for 250
Epochs. The evaluation of the model’s performance is evaluated by calculating the
precision value using 542784 testing Samples.
Disadvantages: Each algorithm has its strengths and weaknesses, and its performance
can depend on factors such as the data used, the context in which they are applied, and
the specific goals of the analysis .
18. Exploring the Vector Space Model for Online Courses :

In this paper the author Roman Shaptala Anna Kyselova, and Gennadiy Kyselov
proposes In this article, we investigate the role of recommender systems and their
potential in the educational system. The key idea is to describe the information
aggregation capabilities of a recommender system to find the best way of tutoring. The
recommender system presented in this article is based on building a vector space model
for massive open online courses (MOOCs). We show that the resulting model can be
used to effectively query semantically related courses as well as search for clusters and

linear transitions between them. Disadvantages: The paper does not specify which
machine learning algorithms are used or how they are evaluated. The paper does not
address the ethical or social implications of using data mining techniques for career
guidance.
19. A New Pedagogy and Online Learning System on Pandemic COVID 19 Era at
Islamic Higher Education:
In this paper the author Triyo Supriyatno, Facrul Kurniawan proposes The purpose
of the article is the analysis of Covid-19 impact on Islamic Higher Education (IHE). The
following research methods were used in the article: empirical methods, methods of
theoretical analysis, questionnaire, and interview. The approaches to online education in
IHE Indonesia have been analyzed. The results of the article have a few significant
implications for the science and practice of online education and its perspectives on the
IHE system. Prospects of online education for students have been offered during Covid-
19. The main recommendations are the following: to organize training courses of online
education methods for lecturers; to organize in-depth training courses of online education
methods for lecturers of non-pedagogical specialties; the university’s management should
provide constant monitoring of the satisfaction of students and lecturers of the online
education organization for the accumulation of statistical data in the dynamics.
or how they are evaluated. The paper does not provide any empirical evidence or user
feedback to support the effectiveness or usability of the proposed system.

20. Online Courses Recommendation System based on Industry Occupation Skills

Requirements:
In this paper, the author Hai-Hui Wang, Chalothon Chootong, Ankhtuya Ochirbat,
Worapot Sommool, W K T M Gunarathn, Timothy K. Shih proposes MOOCs bring us to
a higher education with the concept of flipped classrooms, where students make use of
the online studying materials such as online textbooks, video tutorials, and all sorts of
documents which may take in forms of a web page, online learning platform, educational
learning management systems. We see the astonishing potential of MOOCs in education.
However, there has always been a problem that existed in Taiwan that is also often
discussed. It is known as the gap between industry and education, which means that the
students who have graduated from universities, do not always have the skills that the
industries need. We find that in most cases, students will only have some skills or
knowledge about some tools that are listed from the requirements of the industries. The
students have plentiful self-studying resources from the internet, we hope to encourage
the students to learn and empower themselves by correctly recommending what are the
most required skills of their desired occupation. Therefore, this paper proposed a
clustering method that shows the results of groups of skills that are commonly needed for
a particular type of job.
or how they are evaluated. The paper does not address the ethical or social implications
of using data mining techniques for career guidance

Survey Findings
Our survey aimed to understand the preferences and career aspirations of participants to
tailor course recommendations and predict necessary skills accordingly. The survey was
distributed to a diverse group of individuals across different age groups, educational
backgrounds, and career interests.
Firstly, we analyzed the respondents' career choices. The majority expressed interest
in technology-related fields such as software development, data science, and artificial
intelligence. However, there was also a notable interest in traditional professions like
medicine, law, and education.
Next, we delved int the factors influencing their career decisions. Interestingly, salary
prospects emerged as a significant driver across all age groups, particularly among
younger participants. Job satisfaction, work-life balance, and opportunities for growth
were also cited as crucial considerations.
When it came to selecting courses or educational paths, respondents indicated a

preference for programs that offered practical, hands-on experience and real-world
application of knowledge. Online courses and certifications were favored for their
flexibility and accessibility, especially among working professionals and students
balancing other commitments.
Additionally, we explored the perceived importance of various skills for success in

their chosen careers. Technical skills such as programming, data analysis, and problem-
solving were universally valued, reflecting the increasing demand for STEM-related
expertise. Soft skills like communication, teamwork, and adaptability were also
recognized as essential for career advancement and effective collaboration in professional
settings.

Chapter 3
SOFTWARE REQUIREMENT
SPECIFICATIONS
Stakeholders
Stakeholders of the project are:
 Team Members
 Project Guide
 Customers
 Project Reviewers
 Faculty Department
 College Management
 Organization’s Officials
▪ Customers
▪ Customer Care Representative
▪ Admin
Functional Requirements
For this project, several functional requirements are crucial to ensure its
effectiveness and usability. they are:
Career Assessment and Profiling: The system should provide a comprehensive

career assessment tool that allows users to input their interests, skills, educational
background, and career goals. This profiling process will serve as the foundation for
generating personalized course recommendations and skill predictions.
Course Recommendation Engine: A sophisticated recommendation engine is

needed to analyze the user's profile and match it with relevant courses and educational
programs. This engine should consider factors such as the user's career preferences, skill

gaps, learning style, and preferred mode of study (e.g., online, in-person, self-paced).
Skill Prediction Algorithm: The system must incorporate a robust algorithm to

predict the skills required for success in the user's chosen career path. This algorithm
should leverage industry trends, job market data, and input from professionals in the field to
identify the most in-demand and relevant skills.
Personalized Learning Path: Based on the user's profile and career aspirations, the
system should generate a personalized learning path that outlines the sequence of courses
and activities necessary to acquire the desired skills. This path should be dynamic and
adjustable to accommodate changes in the user's goals or skill development.
Integration with Learning Platforms: The system should integrate seamlessly

with various online learning platforms and educational resources to provide users with
access to recommended courses and materials. This integration should allow for easy
enrollment, progress tracking, and completion certification.
Feedback Mechanism: To enhance user experience and improve the accuracy of

recommendations, the system should incorporate a feedback mechanism. Users should be
able to provide ratings and reviews for recommended courses, as well as feedback on the
relevance and effectiveness of the predicted skills.
User Dashboard and Progress Tracking: A user-friendly dashboard should be

provided to allow users to track their progress, view recommended courses, and monitor
their skill development over time. Progress tracking features such as completion badges,
skill assessments, and performance analytics can help users stay motivated and engaged.
Accessibility and Customization: The system should prioritize accessibility and

customization to accommodate diverse user needs and preferences. This includes support
for multiple languages, adaptive interfaces for different devices, and options for tailoring
recommendations based on specific criteria or constraints.
By fulfilling these functional requirements, this project can effectively empower users to
make informed decisions about their education and career paths, ultimately enhancing their
employability and professional success.

NON-FUNCTIONAL REQUIREMENTS
Non-functional requirements are crucial for ensuring the overall effectiveness,

usability, and performance of this project. These requirements outline the qualities and
characteristics that the system should possess beyond its core functionality. Here are
some key non-functional requirements for this project:
1. Scalability: The system should be able to handle varying loads efficiently,

accommodating an increasing number of users and data without compromising
performance. It should be scalable both vertically and horizontally to adapt to growing
demands seamlessly.
2. Reliability: Users should be able to rely on the system consistently. This entails
minimizing downtime, ensuring data integrity, and implementing robust error handling
mechanisms. The system should recover gracefully from failures and maintain
availability even during maintenance activities.
3. Security: Data security is paramount, especially when dealing with personal

information and career-related data. The system should enforce strict access controls,
encrypt sensitive data, and adhere to industry-standard security practices to prevent
unauthorized access, data breaches, or tampering.
4. Usability: The system should be intuitive and easy to use for users with varying
levels of technical expertise. Clear and concise user interfaces should facilitate smooth
navigation and interaction. Additionally, the system should support accessibility
standards to accommodate users with disabilities.
5. Performance: The system should respond promptly to user requests, delivering

quick recommendations and predictions. Response times should be optimized, and
system latency should be minimized to enhance user satisfaction. Performance metrics
should be monitored and continuously improved upon.

6. Compatibility: The system should be compatible with a wide range of devices,
browsers, and operating systems to ensure accessibility across different platforms. It
should also integrate seamlessly with existing technologies and APIs to leverage external
resources effectively.
7. Maintainability: The system should be designed with maintainability in mind,

making it easy to update, modify, and extend functionalities as needed. Clean and well-
documented code, modular architecture, and version control practices should facilitate
smooth maintenance workflows.
8. Privacy: Users' privacy should be respected and protected throughout their

interactions with the system. Transparent privacy policies should be in place, outlining
how user data is collected, stored, and used. Users should have control over their data and
the option to opt out of data collection if desired.
By addressing these non-functional requirements, this project can ensure that the
system not only meets its functional objectives but also delivers a secure, reliable, and
user-friendly experience for its stakeholders.

System Requirements
Hardware Requirements
A PC with Windows/Linux OS

 Processor with 1.7-2.4gHz speed
 Minimum of 8gb RAM
 2gb Graphic card
Software Requirements
 Text Editor (VS-code/WebStorm)

 Anaconda distribution package (PyCharm Editor)
 Python libraries
1. Anaconda- It is an open source software available to us which enables us to easily codein using
python or R on different operating systems such as the windows, Linux, and Mc OS. It has millions
of users worldwide, and is well known as the industry which helps us in developing systems, testing
them, and training the machines. This further enables us to:
 Manage all the imported libraries, their dependencies, and the environments of developing with
Anaconda.
 In developing techniques to train our machine with TensorFlow, scikit-learn, etc.
 Analyze the datasets and manipulate the with Dask, NumPy, pandas, and Numba
 Visualize or plot the results with Matplotlib, Holoviews, Bokeh, and Datashader.
It also provides us with jupyter notebooks which has all the in-built libraries
embedded in thealready. This eases our coding stress, and also helps us code with
more efficiency
2. Python: The most abundantly used general level programming language. It is used for both a small scale
and big scale systems. It can easily be interpreted. It is said to support multiple programming paradigms. It
includes features of procedural, object-oriented, and functional programming together. It is already garbage-
collected which makes it more efficient.

3. Numpy- It is a python programming library. This basically helps us deal with large datasets,
matrices, and multi-dimensional arrays. It also provides us with a number ofmathematical
functions which help us and ease the calculations. It is an open-source softwareavailable to all.
4. Pandas- It is a library that is written in python language. It helps us with the analyzing of data. It
also provides us with tools and functions to manipulate a large amount of data.
5. Sklearn- It is a library used in machine learning in python programming language. It mainly helps
us with the classification of data, regression of models, and in clustering algorithms. These
algorithms include SVM, random forest etc.
6. Tensorflow- It is a highly know open-source software, which is free and available to all. It is used
for differentiable programming with a large number of tasks. It is used against large data sets
to help us with the dataflow, and its manipulation. It is a basically a math library with various
feature.

Chapter 4
SYSTEM DESIGN AND ANALYSIS
System Analysis
Through the integration of advanced data analytics and machine learning algorithms, this system
aims to analyze career preferences, market trends, and individual skill sets to offer tailored
recommendations for courses and training programs. By leveraging predictive modeling
techniques, it further anticipates future skill demands, empowering users to make informed
decisions about their educational and professional endeavors. This analysis will delve into the
architectural design, functionality, and potential impact of such a system, outlining its role in
facilitating career development and lifelong learning in the digital age.
Fig 4.1 : Activity Diagram

1. Enter Login Information:

 Users begin by providing their login credentials (username and password) to access the movie
streaming platform.
2. Check if Registered User:
 The system checks whether the user is a registered member or a new user.
 If registered, it proceeds to the next step.
 If new, it diverges to a different path.
3. New User Path:
For new users:
 They select movies from the available options.
 The selected movies are added to their favorites list.
4. Registered User Path:
For registered users:
 The system provides a list of movies.
 The user can update their favorites list based on the provided movie options.
 Any changes to the favorites list trigger adjustments in the recommendations list.
5. Update Recommendations:
 The recommendations list is updated based on the user’s actions (adding/removing movies
from favorites).
 This ensures personalized content recommendations for each user.
SYSTEM DESIGN
Systems design is the process of defining the architecture, modules, interfaces, and
data for a system to satisfy specified requirements. Systems design could be seen as the
application of systems theory to product development. There is some overlap with the
disciplines of systems analysis, systems architecture and systems engineering.
The Purpose of this design document is to explore the logical view of architecture design,
data flow diagrams, sequence diagram and an overview of the proposed system for
performing the operations such as signature capturing, pre-processing, feature extraction
and validation which when combined to give the desired output. The design activity
module consist three outputs.
 Architecture Design
 High Level design
 Low Level design
High Level Design
System Architecture
Fig 4.2 : System Architercture
The architecture provides a schematic representation of a learning process within a course

recommendation system utilizing a deep reinforcement learning model. The components
depicted include:
 Learners: Represented by a human figure with a laptop, indicating users of the system.
Information flows from the learners in two forms: Profile Attributes (demographics, in-

terests, past education, etc.) and Followed Courses (courses interacted with).
 Learning Model: A central component responsible for processing input data to generate
course recommendations, part of the larger "Learning Process."
 Deep Reinforcement Learning Model: Represented by a gear icon, suggesting the sys-
tem's use of advanced machine learning techniques, specifically deep reinforcement
learning, to optimize course recommendations.
 Recommended Courses: Depicted by a stack of books at the bottom, indicating the out-
put of the system, i.e., course recommendations generated by the learning model.
 Feedback Loop: Illustrated by green arrows forming a loop, starting from the "Learn-
ers" and looping through "Learners' Feedback" back to the "Learning Process."This
loop allows the system to continually improve recommendations based on user feedback.
 The diagram portrays a dynamic process wherein learner data and feedback iteratively
enhance course recommendations through a deep reinforcement learning model, aiming for
more effective and personalized learning experiences.
Low level Design
Fig 4.3 : Architecture for hybrid approach

1. User ID:
 The process begins with the user providing their unique identification (User ID) to the
recommendation system.
2. Model-Based Collaborative Filter:
 This step represents one of the recommendation techniques used by the system.
 Collaborative filtering involves analyzing user behavior (such as ratings, preferences, and
interactions) to recommend items.
 The model-based approach uses mathematical models to predict user preferences based on
patterns observed in historical data.
 It may involve matrix factorization, latent factor models, or other techniques.
3. Movies:
 The system considers movie-related data, such as movie attributes (genre, director, actors),
ratings, and user interactions.
 Collaborative filtering algorithms compare user preferences with those of similar users to
recommend movies.
4. Content-Based Filter:
 Content-based filtering focuses on the characteristics of items (in this case, movies) rather than
user behavior.
 It analyzes movie features (e.g., genre, plot, keywords) and recommends similar movies based
on these features.
 Content-based filtering is useful when collaborative data is sparse or when personalized
recommendations are needed.
5. Optimal Result:
 The final output is the optimal recommendation for the user.
 It combines the results from both collaborative filtering and content-based filtering.
 The system aims to provide personalized movie recommendations that align with the user’s
preferences.

Data Flow Diagram
Fig 4.4 : DataFlow diagram
1. Load a Data Set:

 The process begins by loading a dataset containing information about movies, users, and their
interactions (e.g., ratings, views).
2. Content-Based Filtering (CBF):
 Content-based filtering focuses on the characteristics of movies themselves.
 It analyzes movie attributes such as genre, director, actors, and plot.
 Based on these features, it generates a list of recommended movies for a particular user.
3. Collaborative-Based Filtering (CF):
 Collaborative filtering considers user behavior and preferences.
 It identifies similar users or items (movies) based on historical data.
 CF generates another list of recommended movies for the same user.
4. Merging Both CBF and CF:
 To enhance recommendation accuracy, the system combines the results from both CBF and
CF.
 This fusion ensures a more comprehensive and personalized set of movie recommendations.
5. Recommending a List of Movies to the User:
 The final output is a curated list of recommended movies tailored to the specific user.
 By leveraging both content-based and collaborative-based approaches, the system aims to

provide relevant and engaging movie suggestions.

Use Case Diagram
Use case diagram is a graph of actors, a set of use cases enclosed by a system boundary,
communication associations between the actor and the use case. The use case diagram describes how
a system interacts with outside actors
Fig 4.5 : Use Case Diagram

Explanation of the workflow of our system:
1) Our system is first divided into training and testing data.
2) The training data is 70% and is given supervised inputs and outputs.
3) The testing data is 30% and shows us how well the system is trained.
4) The dataset we have chosen consists of 13 attributes according to which various algorithms
perform their calculations and approximations.
5) The system starts with first pre processing of the dataset we have fed to it.
6) It studies and analysis it, and then applies the required machine learning algorithm.
7) If it finds that the dataset is supervised, it will separate it into training data and testing data.
8) Otherwise it will stop.
9) The algorithms we are using, are all supervised.
10) After the application of algorithm, internal validation is done .
11) Accuracy is printed in the code itself.
12) Different accuracy is given for different algorithms.
13) We compare the accuracies of all the algorithms and the algorithm that gives the highest
accuracy is the one, which is chosen for prediction, eventually.
14) In our system, the algorithm that gets the highest accuracy rate is DNN.

CHAPTER 5
IMPLEMENTATION
The Proposed System Make Use Different Algorithms and Methods for the
implementation of Hybrid Approach.
Cosine Similarity: Cosine similarity is a measure of similarity between two non-zero

vectors of an inner product space that measures the cosine of the angle between them.
Formula:
Singular Value Decomposition (SVD):Let A be an n*d matrix with singular vectors v1, v2, .
. . , vr and corresponding singular values σ1, σ2, . . . , σr. Then ui = (1/σi )Avi , for i = 1, 2, . .
. , r, are the left singular vectors and by Theorem 1.5, A can be decomposed into a sum of rank
one matrices A.
We first prove a simple lemma stating that two matrices A and B are identical if
Av = Bv for all v. The lemma states that in the abstract, a matrix A can be viewed as a
transformation that maps vector v onto Av.

Methodology
Deep learning: (also known as deep structured learning) is part of a broader family
of machine learning methods based on artificial neural networks with representation
learning. Learning can be supervised, semi-supervised or unsupervised.[1][2][3]
Deep learning architectures such as deep neural networks, deep belief

networks, recurrent neural networks and convolutional neural networks have been applied
to fields including computer vision, speech recognition, natural language
processing, audiorecognition,socialnetwork filtering, machine translation, bioinformatics,
drug design, medical image analysis, material inspection and board game programs,
where they have produced results comparable to and in some cases surpassing human
expert performance.
Artificial neural networks (ANNs) were inspired by information processing and

distributed communication nodes in biological systems. ANNs have various differences
from biological brains. Specifically, neural networks tend to be static and symbolic, while
the biological brain of most living organisms is dynamic (plastic) and analog.
The adjective "deep" in deep learning comes from the use of multiple layers in the
network. Early work showed that a linear perceptron cannot be a universal classifier, and
then that a network with a nonpolynomial activation function with one hidden layer of
unbounded width can on the other hand so be. Deep learning is a modern variation which is
concerned with an unbounded number of layers of bounded size, which permits practical
application and optimized implementation, while retaining theoretical universality under
mild conditions. In deep learning the layers are also permitted to be heterogeneous and to
deviate widely from biologically informed connectionist models, for the sake of
efficiency, trainability and understandability, whence the "structured" part.

Front End Technologies
In react.js we used axios npm module to fetch the data from the api that is generated from
flask. The front-end technology stack plays a pivotal role in shaping the user experience and
facilitating seamless interaction with the system. At its core, HTML (HyperText Markup
Language) forms the foundation of the front-end, providing the structural framework for web
pages. It delineates the layout of content elements, ensuring clarity and organization in
presenting information to users.
Complementing HTML, CSS (Cascading Style Sheets) comes into play to enhance the visual
appeal and aesthetics of the user interface. CSS allows for the customization of fonts, colors,
layouts, and other stylistic attributes, thereby creating a cohesive and visually pleasing design
that aligns with modern web standards and user expectations.
Fig 5.1 : Sample code snippet of React.js

Driving interactivity and dynamic functionality, JavaScript emerges as a crucial component of

the front-end technology stack. JavaScript empowers developers to incorporate interactive
elements, handle user inputs, perform client-side validation, and facilitate real-time updates
without the need for page reloads. Its versatility makes it indispensable for creating engaging
user experiences and responsive interfaces.
In the realm of front-end frameworks, options like React.js or Vue.js offer powerful tools for
building interactive and component-based UIs. These frameworks streamline the development
process by promoting code reusability, efficient state management, and seamless integration
with other libraries or APIs. By adopting such frameworks, developers can accelerate the pace
of development while maintaining a high level of flexibility and scalability.
Moreover, for ensuring consistency and responsiveness across different devices and screen
sizes, front-end frameworks like Bootstrap or Material-UI provide a rich set of pre-designed
UI components and responsive layout systems. These frameworks empower developers to
create adaptive and visually appealing interfaces that deliver a seamless user experience
regardless of the user's device.
In summary, the front-end technology stack for this project encompasses HTML, CSS,
JavaScript, and relevant frameworks/libraries. Together, these technologies enable the
creation of an intuitive, visually appealing, and interactive user interface that empowers users
to explore career options, receive personalized recommendations, and gain insights into future
skill requirements with ease and efficiency.

Back End Technologies
For backend we have use flask app to generate a local host api the resultant api is fetch in front to display
the result. We have developed our machine learning model in python .
By using flask, we generate resulting api which stores the data in the form of json format these
data is retrieved in react by using axios npm mode and then displaying the data
The choice of back-end technology is crucial in ensuring the efficient processing of data,
managing user accounts, and delivering personalized recommendations. At the heart of the
back end lies the server-side programming language, which handles the logic and functionality
behind the scenes.
Fig 5.2 : Backend code snippet

One popular option for server-side development is Python, known for its simplicity,
versatility, and extensive ecosystem of libraries and frameworks. Frameworks like Django or
Flask provide robust tools for building scalable web applications, handling user
authentication, and interfacing with databases, which are essential components for a system
focused on course recommendation and skill prediction.
Another noteworthy technology in the back-end arsenal is the use of databases to store and
manage data efficiently. Relational databases like PostgreSQL or MySQL are commonly
employed for their ability to organize data into structured tables and perform complex queries.
Additionally, NoSQL databases such as MongoDB offer flexibility in handling unstructured
or semi-structured data, which could be beneficial for accommodating diverse user profiles
and preferences.
In the context of machine learning and predictive analytics, libraries like TensorFlow or
scikit-learn in Python provide powerful tools for building and deploying predictive models.
These libraries enable the system to analyze user data, predict future skill demands, and
generate personalized course recommendations based on career choices and individual
profiles.
Moreover, the back end often involves integrating with external APIs or services to enrich the
system's functionality. APIs provided by job boards, educational platforms, or professional
networks can supply valuable data on job listings, course offerings, and industry trends, which
can inform the recommendation and prediction algorithms.
Finally, considerations for scalability, security, and performance are paramount in the
selection of back-end technologies. Cloud platforms like Amazon Web Services (AWS),
Microsoft Azure, or Google Cloud Platform (GCP) offer a range of services for hosting,
scaling, and securing web applications, ensuring reliability and responsiveness even under
heavy user loads. Together, these technologies form a robust foundation for building a
dynamic and intelligent system that empowers users in their career development journey.

Chapter 6
TESTING
Design of the Test Cases
Testing is the process of evaluating a system or its components with the intent to
find that whether it satisfies the specific requirement or not. This activity results in actual,
expected and the difference between their results. In simple words testing is executing a
system in order to identify any gaps, errors or missing requirements in contrary to the actual
desire or requirements.
According to the ANSI/IEEE 1059 standard, Testing can be defined as “A process

of analyzing a software item to detect the differences between existing and required
conditions and to evaluate the features of the software items”.
Testing is the practice of making objective judgments regarding the extent to which
the system (device) meets, exceeds or fails to meet stated objectives.
A good testing program is a tool for agency and integrator supplier it identifies the
end of the “Development” phase of the project, establishes the criteria for project
acceptance, and establishes the start of the warranty period.
Types of Test
There are many types of testing which can be carried out. Few of the testing types are: 1.
Unit testing 2. Integration testing 3. Functional testing 4. System Testing
Unit Testing: Unit testing involves the design of test cases that validate that the internal
program logic is functioning properly, and that program inputs produce valid outputs. All
decision branches and internal code flow should be validated. It is the testing of individual
software units of the application.
It is done after the completion of an individual unit before integration. This is a structural
testing, that relies on knowledge of its construction and is invasive. Unit tests perform basic
tests at component level and test a specific business process, application, and/or system

configuration. Unit tests ensure that each unique path of a business process performs
accurately to the documented specifications and contains clearly defined inputs and
expected results. Unit testing is usually conducted as part of a combined code and unit test
phase of the software lifecycle, although it is not uncommon for coding and unit testing to
be conducted as two distinct phases.
Integration Testing: Integration tests are designed to test integrated software components
to determine if they actually run as one program. Testing is event driven and is more
concerned with the basic outcome of screens or fields. Integration tests demonstrate that
although the components were individually satisfaction, as shown by successfully unit
testing, the combination of components is correct and consistent. Integration testing is
specifically aimed at exposing the problems that arise from the combination of
components. Software integration testing is the incremental integration testing of two or
more integrated software components on a single platform to produce failures caused by
interface defects. The task of the integration test is to check that components or software
applications, e.g. components in a software system or – one step up – software applications
at the company level – interact without error.
Acceptance Testing: User Acceptance Testing is a critical phase of any project and
requires significant participation by the end user. It also ensures that the system meets the
functional requirements.
Functional Testing: Functional tests provide systematic demonstrations that functions

tested are available as specified by the business and technical requirements, system
documentation, and user manuals.
Functional testing is centered on the following items:
Valid Input : identified classes of valid input must be accepted.
Invalid Input : identified classes of invalid input must be rejected.
Functions : identified functions must be exercised.
Output : identified classes of application outputs must be exercised.
Systems/Procedures : interfacing systems or procedures must be invoked.

Organization and preparation of functional tests is focused on requirements, key

functions, or special test cases. In addition, systematic coverage pertaining to identify
Business process flows; data fields, predefined processes, and successive processes must
be considered for testing. Before functional testing is complete, additional tests are
identified and the effective value of current tests is determined.
System Testing: System testing is actually a series of different tests whose primary
purpose is to fully exercise the computer-based system. Although each test has a different
purpose, all work to verify that all the system elements have been properly integrated and
perform allocated functions. The testing process is actually carried out to make sure that
the product exactly does the same thing what is supposed to do. In the testing stage
following goals are tried to achieve: -
● To affirm the quality of the project.

● To find and eliminate any residual errors from previous stages.
● To validate the software as a solution to the original problem.
Test strategy and approach

Field testing will be performed manually and functional tests will be written in detail
Test objectives
• All field entries must work properly.
• Pages must be activated from the identified link.
• The entry screen, messages and responses must not be delayed.
Features to be tested
• Verify that the entries are of the correct format
• No duplicate entries should be allowed
• All links should take the user to the correct page.

RESULTS AND DISCUSSION
Since our project is Online course recommendation system .one can develop a Online
course recommendation system by using either content based or collaborative filtering or
combining both.
In our project we have developed a hybrid approach i.e combination of both content and
collaborative filtering .Both the approaches have advantages and dis-advantages
.in content based filtering the it based on the user ratings or user likes only such kind of
Online course will recommended to the user.
Advantages: it is easy to design and it takes less time to compute
Dis-advantages: the model can only make recommendations based on existing interests of
the user. In other words, the model has limited ability to expand on the users' existing
interests.
In Collaborative filtering the recommendation is comparison of similar users.
Advantages: No need domain knowledge because the embeddings are automatically

learned. The model can help users discover new interests. In isolation, the ML system may
not know the user is interested in a given item, but the model might still recommend it
because similar users are interested in that item.
Dis-advantages: The prediction of the model for a given (user, item) pair is the dot product
of the corresponding embeddings. So, if an item is not seen during training, the system can't
create an embedding for it and can't query the model with this item. This issue is often
called the cold-start problem.
The hybrid approach will resolves all these limitations by combining both content and collaborative
filtering

Fig 6.1 : Comparison between the three approaches

SNAPSHOTS
Snapshot 7.1:Selection of Plant

Snapshot 7.3: Uploading image
Snapshot 7.4:Levels of classification

Snapshot 7.5: Description of disease
Snapshot 7.6: Remedies

Snapshot 7.7: Healthy leaf classification
Snapshot 7.8 :Classification of untrained classes

FUTURE ENHANCEMENT
In the proposed approach, It has considered Genres of Online courses but, in future we can
also consider age of user as according to the age Online course preferences also changes,
like for example, during our childhood we like animated Online courses more as compared
to other Online courses. There is a need to work on the memory requirements of the
proposed approach in the future. The proposed approach has been implemented here on
different Online course datasets only. It can also be implemented on the Film Affinity and
Netflix datasets and the performance can be computed in the future.

CONCLUSION
In this project, to improve the accuracy, quality and scalability of Online course
recommendation system, a Hybrid approach by unifying content based filtering and
collaborative filtering; using Singular Value Decomposition (SVD) as a classifier and
Cosine Similarity is presented in the proposed methodology. Existing pure approaches and
proposed hybrid approach is implemented on three different Online course datasets and the
results are compared among them. Comparative results depicts that the proposed approach
shows an improvement in the accuracy, quality and scalability of the Online course
recommendation system than the pure approaches. Also, computing time of the proposed
approach is lesser than the other two pure approaches.

BIBILOGRAPHY
[1] K. K. Chandriah and R. V. Naraganahalli, “RNN/LSTM with modified Adam
optimizer in deep learning approach for auto◻mobile spare parts demand
forecasting,” Multimedia Tools and Applications, vol. 80, no. 17, pp. 26145–26159,
2021.
[2] S. Nagrecha, J. Z. Dillon, and N. V. Chawla, “MOOC dropout prediction: lessons

learned from making pipelines interpret◻able,” in Proceedings of the 26th
International Conference on World Wide Web Companion, pp. 351–359, NY, ACM,
April 2017.
[3] WideResNet with Joint Representation Learning and Data Augmentation for
Course recommendation Identification Shichao Hu1, Bin Zhang1 , Jinhong Lu2 ,
Yiliang Jiang1 , Wucheng Wang1 , Lingcheng Kong1 , Weifeng Zhao1 , Tao Jiang.
[4] Ensemble-based course recommendation; Marc Sarfati, Anthony Hu, Jonathan

Donier
[5] MulKINet: Multi-Stage Key-Invariant Convolutional Neural Networks for

Accurate and Fast Skill Prediction Identification by Chengdi Cao; Wei-Qiang Zhan.
[6] Chuang AC, Huang NF, Tzeng JW, Lee CA, Huang YX, Huang HH (2021)
MOOCERS: Exercise recommender system in MOOCs based on reinforcement
learning algorithm. In: 8th International Conference on Soft Computing & Machine
Intelli◻gence (ISCMI), Cario, Egypt, 2021, pp 186-190.6. Y. Yang, Y. Zhu, and Y.
Li, “Personalized recommendation with knowledge graph via dual#autoencoder,”
Applied Intelli◻gence, vol. 52, no. 6, pp. 6196–6207, 2022.
[7] A. N. Ngaffo, W. E. Ayeb, and Z. Choukair, “Service recommendation driven by

a matrix factorization model and time series forecasting,” Applied Intelligence, vol.
52, no. 1, pp. 1110–1125, 2022.

[8] Bellman R (2013) Dynamic programming. Dover Publications.
[9] Wang M, Peng J, Cheng B, Zhou H, Liu J (2011) Knowledge visualization for
self#regulated learning. EducTechnol Soc 14(3):28–42.
[10] Chuang AC, Huang NF, Tzeng JW, Lee CA, Huang YX, Huang HH (2021)
MOOCERS: Exercise recommender system in MOOCs based on reinforcement
learning algorithm. In: 8th International Conference on Soft Computing & Machine
Intelli◻gence (ISCMI), Cario, Egypt, 2021, pp 186-190.
[11] Micarelli A, Stamper J, Panourgia K (eds) (2016) Springer,Cham, pp 267–272.
[12] Nemoto T, Beglar D (2014) Likert-scale questionnaires. In: JALT 2013

conference proceedings, pp 1–8.
[13] Patel PG, Carver N, Rahimi S (2011) Tuning computer gaming agents using
q#learning. In: 2011 federated conference on com◻puter science and information
systems (FedCSIS). IEEE, pp 581–588.
[14] Park C, Kim D, Oh J, Yu H (2016) Improving top-k recommendation with

truster and trustee relationship in user trust net◻work. Inf Sci 374:100–114.
[15] Wang Y, Wang L, Li Y, He D, Chen W, Liu T-Y (2013) A theoretical analysis

of NDCG ranking measures. In: Proceedings of the 26th annual conference on
learning theory (COLT 2013),vol 8. Citeseer, p 6.
[16] Onah DFO, Pang ELL, Sinclair JE (2022) An investigation of self-regulated

learning in a novel MOOC platform. J Comput Higher Educ 25:1–34.
[17] Imran K, Abdullah N (2009) Building an e-learning recommender system using

vector space model and good learners average rating, pp 194–196.
[18] Taghipour N, Kardan A (2008) A hybrid web recommender system based on

q#learning. In: Proceedings of the 2008 ACM symposium on applied computing.

[19] Kober J, Bagnell JA, Peters J (2013) Reinforcement learning in robotics: a
survey. Int J Robot Res 32(11):1238–1274.
[20] Wei X, Saab N, Admiraal W (2023) Do learners share the same perceived
learning outcomes in MOOCs? Identifying the role of motivation, perceived learning
support, learning engagement, and self-regulated learning strategies. Internet Higher
Educ 56:100880

PLANT LEAF DISEASE DETECTION

Smithashree KP1, Bhoomika M Rao2, Prerana N3, Rahul R4, Spoorthi
Ravish5 1Asst.Professor, Dept of ISE, Maharaja Institute of
TechnologyMysore, India.
2,3,4,5
Engineering Student, Dept of ISE, MITM
get. Therefore, for the better yield in agriculture, detection of

Abstract: Early detection of plant disease is important for
disease in plants plays an instrumental role. Plants are highly
agriculture for the effective production and also to improve
prone to diseases that affect the growth of the plant which in
the economy of the country. Automatic detection and
turn affects the profit of the farmer. In order to detect a plant
classification of plant disease and providing required
disease at very initial stage and to avoid its spread, use of
remedy proves to be one of the efficient technique for
agriculture. This paper presents one of the computer vision
technique, a convolutional neural network(CNN) with the
transfer learning method for effective classification of
diseases in 3 crops namely Capsicum, Potato and
Strawberry. The above technique effectively provides an
accuracy of 95%.
Keywords: Convolutional Neural Network(CNN),
Computer vision, classification, transfer learning.
I. INTRODUCTION
Agriculture is the art and science of cultivating plants and
livestock. It is one of the major sectors of the Indian
economy as it contributes about 18% to the total GDP and
provides employment to over 60% of the population and is
present in the country for thousands of years. Over 75 per
cent of the rural households depend on agriculture.
Nowadays there is a tremendous loss in quality and quantity
of food crops yield because of various diseases affecting the
plant. For this purpose proper steps should be taken to rescue
the plants from disease and to increase the yield. Early
detection of these diseases can allow to take preventive
measures and mitigate economic and production losses. A
leaf is one of the most important part of a plant responsible
for photosynthesis. So, detecting the disease through leaf
will be one of the primary sectors in finding the disease.
So the objective of this work is to develop a software, that is
effective and error-free disease detection system for plant.
To overcome the leaf disease, a software in been developed
particularly for three species i.e., Potato, Capsicum and
Strawberry which automatically detect the disease using the
image of the leaf and also provide the remedies for those
diseases
II. PROBLEM STATEMENT

One of the important sectors of Indian Economy is
Agriculture. Employment to almost 55% of the countries
workforce is provided by Indian agriculture sector. India is
one of the largest producer of pulses, rice, wheat, spices and
spice products. Farmer's economic growth mostly depends
on the quality of the products that they produce, which in
turn mostly relies on the plant's growth and the yield they

automatic disease detection technique is advantageous. The
symptoms of plant diseases are commonly found on the leaf
of the plants. Manual detection of plant disease using leaf
images with the help of experts is a tedious job. Hence, it is
required to develop computational methods which will
make the process of disease detection and classification
using leaf images automatically for better yield and profit
for the farmers.
III. EXISTING SYSTEM

In India, farmers still uses the traditional method to detect
the diseases in the leaf. Farmers or experts uses their naked
eye observation to find the disease .This how the
identification and detection of leaf diseases is done. So in
order to do this work, a team of knowledge person as well
as a continuous monitoring of plant is necessary, which
costs very high when we do with large farms and it can be
done to the limited area. But in few countries, farmers do
not have proper facilities or even idea that they could
contact to experts. And also consulting experts even isn't a
cost friendly as well as it is not time consuming. In such
conditions, Automatic detection of the diseases by just
seeing the symptoms on the plant leaves makes it easier as
well as cheaper. Whereas if automatic detection technique
is used it will take less efforts, less time and gives more
accuracy. In plants, some general diseases seen are brown
and yellow spots, early and late scorch, and others are
fungal, viral and bacterial diseases. Image processing is
used for measuring affected area of disease and to
determine the affected area.
IV. PROPOSED SYSTEM

The proposed system makes use of one of the deep learning
technique called the Convolutional Neural Network(CNN).
CNN are mainly used for classification problems. The
proposed system involves transfer learning methodology
i.e., using one of the pre-trained architecture to get better
results. VGG19 is used as the transfer learning architecture
which is composed of 19 deep layers. The system takes in
an image of a leaf as an input through the device camera or
its internal storage and goes through three level of
classification described below:
 Healthy/Unhealthy(proceeds further if classified
as unhealthy)
 Classification of the Crop
 Disease classification
Once the disease is been finally classified its related
description about the disease and its corresponding
remedies are provided in order to prevent the further spread
of the disease. The proposed system is capable of
identifying the diseases such as pepper bell bacterial spot,
potato early

blight, potato late blight, strawberry leaf scorch including

their healthy images of each crops.
IV. METHODOLOGY
Convolutional Neural Network: (CNN) is a Deep Learning
algorithm which can take in an input image, and process
them to classify the given image. The pre-processing required
in a CNN is much lower as compared to other algorithms and
Image Processing techniques. CNN is capable of applying
many filters automatically It mainly consist of four layers:
 Convolution layer: In this layer series of
mathematical operations are performed to extract the VGG19 Architecture
feature map of the input image. We have used plant village dataset consisting of 3 crops
namely Pepper bell, Strawberry, Potato. The dataset is
 Pooling layer: It is mainly used to reduce the size of
divided into 80% train and 20%test.5124 images were used
the output matrix from the convolution layer. There
to train the model and 1226 images for testing. The images
are 3 different types of pooling: Max, Min, Average
pooling. selected from dataset is cropped to the size of 224X224 each
input image matrix is been convoluted and reLU activation
 Flattening layer: is converting the data into a 1-
function is been applied. The fully connected of VGG19 is
dimensional array for inputting it to the next layer.
modified as per our project requirements. The fully
The output from the convolutional layers are
connected layer is composed of 2 hidden layers consisting of
flattened to get a single feature vector.
512 and 256 neurons respectively, Adam is used as optimizer
 Fully connected layer: The output from the as it converges fast and a batch size of 32 was used during
flattening layer is fed into the fully connected layer. training of the model. The output layer is multi neuron
Recognition and classification is performed in this output which gives out the final results. The system consists
layer. of 3 levels of classification. At the first stage the given leaf
input is classified to be healthy or unhealthy, if the result is
The system uses transfer learning technique i.e., VGG19
classified at healthy the further processing is not done else
model has described below
the image is passed to the second level of classification
 VGG19 is a variant of VGG model which in short where the type of crop is identified at the last stage of
consists of 19 layers that includes 16 convolution classification. The final disease have been outputted. Based
layers, 3 Fully connected layer, 5 MaxPool layers on the type of disease identified the remedies for the same is
and 1 SoftMax layer. There are other variants of given out.
VGG like VGG11, VGG16 and others.
 A fixed size of (224 * 224) RGB image was given
as input to this network which means that the matrix
was of shape (224,224,3).
 They subtracted the mean RGB value from each
pixel, which was the only preprocessing which was
done and computed over the whole training set.
 Used kernels of (3 * 3) size with a stride size of 1
pixel, this enabled them to cover the whole notion of
the image.
 Spatial padding was used to preserve the spatial
resolution of the image.
 Max pooling was performed over a 2 * 2 pixel
windows with stride 2. Data flow diagram
 This was followed by Rectified linear unit(ReLu) to
introduce non-linearity to make the model classify
better and to improve computational time as the
previous
 models used tanh or sigmoid functions this proved
much better than those.
 It was constructed with three fully connected layers
from which first two were of size 4096 and the next
layer with 1000 channels for 1000-way ILSVRC
classification and the final layer is a softmax
Dept of IS&E, MITM
function. 54
System Architecture
V. EXPERIMENTAL RESULTS further hence, we have come with the system that
The snapshots of the results are shown below. The system is automatically detects the disease affected to a plant with the
capable of detecting the diseases with 95% accuracy. If the help of the leaf image. The symptoms of the disease appears
system is given with the crop that is not trained it will show a on the leaf region and hence taken as an input for further
message stating that the given crop was not trained. The detection and classification. These images are classified
Fig(a) shows the three level classification for an given potato using CNN and transfer learning methods by using VGG19
leaf as input. Fig(b) shows the description and remedy architecture. The system is capable of finding disease with
associated with that disease. Fig(c) shows first level of 95% accuracy. And shows three level classification if it is
classification since it is a healthy image. detected as unhealthy and one level of classification if it’s
healthy.
VII. FUTURE SCOPE

In future the same technique can be extended to multiple
crops to detect more number of diseases. Also many other
deep learning technique can be used to enhance the accuracy.
More number of dataset can be used to get better results.
ACKNOWLEDGEMENT
We are indeed grateful to many groups of people of who
have helped us with various aspects of this study. We would
like to thank Prof Smithashree K P as well as special thanks
for our head of the department Dr Sharath Kumar Y H for
providing us the opportunity to do this amazing project on
“Plant leaf disease detection”.
Fig(a)
REFERENCES
[1] Melike Sardogan, Adem Tuncer, Yunus
Ozen, Plant Leaf Disease Detection and
Classification based on CNN with LVQ
Algorithm, 3rd International
Conference on Computer Science and
Engineering, IEEE, 2018.
[2] Jia Shijie, Jia Peiyi, Hu Siping, Liu
Haibo, Automatic Detection of Tomato
Diseases and Pests Based on Leaf
Images, 2017, IEEE.
[3] Eftekhar Hossain, Md. Farhad Hossain
and Mohammad Anisur Rahaman, A
Color and Texture Based Approach for
the Detection and Classification of
Plant Leaf Disease Using KNN
Classifier, 2019 International
Conference on Electrical, Computer and
Fig(b) Communication Engineering (ECCE),
7-9 February, 2019.
[4] Neha G. Kurale, Madhav V. Vaidya,
Classification of Leaf disease using
Texture Feature and Neural Network
Classifier, IEEE, 2018.
[5] Trimi Neha Tete, Sushma Kamlu,
Detection of Plant Disease Using
Threshold, K- Mean Cluster and ANN
Algorithm, 2017 2nd International
Conference for Convergence in
Fig(c) Technology (I2CT), IEEE.
[6] Robert G. de Luna, Elmer P. Dadios,
VI. CONCLUSION Argel Bandala, Automated Image
Dept
Detection of of
theIS&E,
plant MITM
disease at its early stages proves to be 55
Capturing System for Deep Learning-
one of the best method to avoid the spread of the disease based Tomato Plant Leaf Disease
Detection and Recognition, Proceedings
of TENCON 2018 - 2018 IEEE Region
10 Conference
(Jeju, Korea, 28-31 October 2018).

Final Report 1

Uploaded by

Copyright:

Available Formats

Final Report 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Final Report 1

Uploaded by

Copyright:

Available Formats

VISVESVARAYA TECHNOLOGICAL UNIVERSITY

“JNANA SANGAMA”, BELAGAVI, KARNATAKA-590018

“ COURSE RECOMMENDATION AND SKILL PREDICTION

DEPARTMENT OF INFORMATION SCIENCE AND ENGINEERING

MAHARAJA INSTITUTE OF TECHNOLOGY, MYSORE

BELAWADI, NAGUVANAHALLY POST, S.R. PATNA TALUK,

MAHARAJA INSTITUTE OF TECHNOLOGY MYSORE

Signature of guide Signature of the Principal Signature of the

Name of the Examiners External Viva Signature with date

We are greatly indebted to our guide Prof.Amruth v, Asst. Professor, Department of

We are obliged to all teaching and non-teaching staff members of Department of

The primary objective of an Online Course Recommendation System is to enhance the

Chapter 1 : INTRODUCTION Page-no

Chapter 2 : LITERATURE SURVEY

Chapter 3 : SYSTEM REQUIREMENT SPECIFICATION

Chapter 4 : SYSTEM ANALYSIS AND DESIGN

SNAPSHOTS AND RESULT DISCUSSION 43

Dept of IS&E, MITM 1

Dept of IS&E, MITM 2

 System developed as real time application meant for the companies.

 System uses both supervised learning algorithms and unsupervised learning

Dept of IS&E, MITM 3

The following papers/thesis/journals are referred for information regarding our

2. Online Course Recommendation Using Deep Convolutional Neural Network with

Dept of IS&E, MITM 4

3. A Job Recommendation Method Optimized by Position Descriptions and

Dept of IS&E, MITM 5

4. An Intelligent Career Guidance System using Machine Learning:

5. Exploring Social Media Data for MOOC Recommendation:

Dept of IS&E, MITM 6

6. Educational Data Analytics using Association Rule Mining and Classification:

7. Engineering Students’ Job Performance Prediction Model based on Adversity Quotient

Dept of IS&E, MITM 7

8. Supporting Users of Open Online Courses with Recommendations: an

Dept of IS&E, MITM 8

9. HYBRID MASSIVE OPEN ONLINE COURSE (MOOC)

Dept of IS&E, MITM 9

11. Smart Career Guidance and Recommendation System:

In this paper the author Lakshmi Prasanna, DR.D.Haritha proposes Recommender

Dept of IS&E, MITM 10

12. A Study of the Mass Customization-based Strategy for the Recommendation of

Disadvantages: The system is designed to recommend courses based on similarities

13. A Collaborative Recommendation system for online course recommendations:

Dept of IS&E, MITM 11

15. Research on Optimal Design of Online Education Course Recommendation

16. Massive open online course recommendation system based on a reinforcement

Dept of IS&E, MITM 13

In this paper, the author Siriporn Sakboonyarat, Panjai Tantatsanawong proposes

18. Exploring the Vector Space Model for Online Courses :

Dept of IS&E, MITM 14

Dept of IS&E, MITM 15

20. Online Courses Recommendation System based on Industry Occupation Skills

Dept of IS&E, MITM 16

When it came to selecting courses or educational paths, respondents indicated a

Additionally, we explored the perceived importance of various skills for success in

Dept of IS&E, MITM 17

Career Assessment and Profiling: The system should provide a comprehensive

Course Recommendation Engine: A sophisticated recommendation engine is

Dept of IS&E, MITM 18

Skill Prediction Algorithm: The system must incorporate a robust algorithm to