Proceedings of the SMART–2022, IEEE Conference ID: 55829

11th International Conference on System Modeling & Advancement in Research Trends, 16th–17th, December, 2022
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India

Analyzing the Student’s Risk in using

Electronic Gadgets using Hybrid Machine
2022 11th International Conference on System Modeling & Advancement in Research Trends (SMART) | 978-1-6654-8734-4/22/$31.00 ©2022 IEEE | DOI: 10.1109/SMART55829.2022.10047035

Learning Model as Case Study

V. Belsini Gladshiya1 and Dr. K. Sharmila2

Research Scholar, Dept. of Computer Science, VISTAS,
1

Assistant Professor, Department of Computer Science, Agurchand Manmull Jain College, Chennai, India
2
Research Supervisor & Associate Professor, Department of Computer Science, (VISTAS), Chennai
Email: [email protected], [email protected]
Abstract—An exhaustive deformation in the field of II. Predictive Analytics
education made a mark in holding the electronic gadgets
An advanced branch of data analytics which predicts
in the hands of the students for learning today due to the
pandemic situation of Covid’19. Since education is the
the future probability of occurrences based on the historic
elemental process of every human, the educational institutions data present. It consists of various statistical and analytical
and the government put forward the learning procedure methods to create a Predictive analytical model.[1]
through virtual mode of learning. This made the chance of The prediction can be done by the steps followed
operating the electronic gadgets by the student community in 1. Collecting the data from a data source.
a numerous way than the adults. Data analytics is the process 2. Pre-process the data.
of analyzing the data to infer certain predictions and patterns 3. Transformation of data.
to make decisions in a better way in future. For data analytics 4. Using of Predictive models to analyse the data.
the machine learning algorithms are the essential methods of
5. Prediction of the results according to the data.
handling data in predictive analytics. The machine learning
algorithms are combined to form a hybrid approach for data III. Hybrid Machine Learning:
analytics to improve the accuracy of the model. This paper Machine Learning is an important concept in artificial
reveals the hybrid approach of machine learning model to intelligence which provide the essential of data for the
analyse the student data to analyse the risk of the students
machine to train the data and exposed the training data to
using the electronic gadgets during online learning.
test the new data to predict the result.[1]
Keywords: Predictive Analytics, Hybrid Machine Learning, The hybrid machine learning algorithms are the
Ensemble Techniques, Methodology, Risk Analysis
combination of two different machine learning techniques
I. Introduction which improves the classification and prediction
performance of the algorithms. This hybrid approach of
Education is essential for every child for their life .Due machine learning algorithms brings better performance
to the pandemic situation the learning process is converted and the model becomes more extensive support for data
into online learning and the students from the primary classes analytics.[2]
to the graduates are well in advance of using the electronic
IV. Ensemble Methods
gadgets like smart phones, laptops, and computers. Even
though the learning process continued in online there are Ensemble Methods are techniques in machine learning
which combines the basic model of algorithms to produce
some disadvantages or difficulties in the online learning
an advanced machine learning models.( Evan Lutins, 2017).
process. The students are carrying the gadgets instead of
carrying the books and notes. Using the gadgets they have A. Types of Ensemble Methods
to face plenty of disturbances, distractions and disabilities in 1) Voting Classifier
their health and mind. They were addicted in playing online The voting classifier is a technique in which the
games and spending time in social media than spending machine learning models are grouped based on highest
the time with their family and friends. This paper briefly voting probability of the classes selected and predict the
accounts the risk factors of the students using the gadgets result according to that. It is mainly used in classification
by analysing with the hybrid machine learning algorithms problems.
and their results were discussed to overcome this situation 2) Stacking
to deliver the younger generation of today who will be the Stacking is also an ensemble technique which uses
leaders of tomorrow. the predictions from various models to predict the final

Copyright©IEEE–2022 ISBN: 978-1-6654-8734-4 | 689

Authorized licensed use limited to: Sri Sai Ram Engineering College. Downloaded on November 19,2024 at 08:40:02 UTC from IEEE Xplore. Restrictions apply.
 11th International Conference on System Modeling & Advancement in Research Trends, 16th–17th, December, 2022
College of Computing Sciences & Information Technology, Teerthanker Mahaveer University, Moradabad, India

prediction by building a new model. This model works B. Data Set:

based on a meta learner to lower the weakness and improves The real data set is collected from the parents, students
the strength of each model to make the performance better. and teachers about the usage of electronic gadgets from the
[4] primary school to the graduates through Google forms. The
3) Boosting data set consists of 600 rows and 16 columns.
Boosting is an ensemble method generally used to
build a strong classifier from the weak classifiers. This can
be performed by developing a model containing the training
data set and the new model is made as the corrected model
than the first models.
V. Related Works
The hybrid model is predicting the student employment
after their graduation. This helps the administration to
identify students of poor academic performance and low
employability and to take timely actions. (inutha K, Yogisha
H K).[5]
Predicting academic performance of engineering
students using ensemble method. (tek bist bithari, sharan
thapa, and hari k.c., 2020 )[6] Fig. 2: Data Set
Different classifiers such as decision tree, K-NN, Naïve
Attributes used are
Bayes approach, Random Forest naïve Bayes resulted with
Table 1. Attributes used for Analysis
a highest accuracy of 89% and Random forest resulted with
a lowest accuracy of 69% for employability prediction.
(Hugo, Linsey)[7].
The student academics is predicted using a majority
vote based ensemble classifier (A. Salini, U. Jeyapriya,
2018)[8].
Predicting employment prior to graduation, the
research used commonly recognized and advanced machine
learning models, decision trees, neural networks and
logistic regression. Results demonstrate that employment
prior to graduation can be predicted with73% accuracy with
a neural network as the most accurate predictive model.
(Neeraj Khadilkar and Deepali Joshi)[9]
VI. Methodology – Hybrid Machine Learning
–Ensemble Voting Classification Model
In this research a hybrid machine learning model C. Data Pre-processing
is developed using the ensemble voting classification
The data is pre-processed to add or delete the rows
techniques and methods. This hybrid model combines
or columns and it is also checked there are null values or
the machine learning algorithms Logistic regression and
not. It is identified that no null value is present.
Support vector machine by means of voting classification.
A. Work Process

Fig. 1: Work Process of Hybrid Machine Learning Model Fig. 3: Missing Value Detection

Copyright©IEEE–2022 ISBN: 978-1-6654-8734-4 | 690

Authorized licensed use limited to: Sri Sai Ram Engineering College. Downloaded on November 19,2024 at 08:40:02 UTC from IEEE Xplore. Restrictions apply.
 Analyzing the Student’s Risk in using Electronic Gadgets using Hybrid Machine Learning Model as Case Study

D. Data Transformation identify the risk of the students using the electronic gadgets
Data transformation is done to convert categorical data, for their learning. The hybrid voting classifier model
or text data, into numbers, which our predictive models performs better than Logistic algorithm and SVM. The
can better understand. Label Encoding is the technique to future research can be done to increase the accuracy of the
change the labels into numeric form so as to convert it into model. Some more metrics can be tested and its accuracy
the machine-readable form. can be identified. The student data sets from other aspects of
difficulties can be identified and their success probabilities
E. Data Analytics
can be predicted. Various other data sets such as analyzing
The data set is processed using Logistic regression the risk of using gadgets for their profession in IT field and
and Support Vector machine individually and then it is BPO’s to predict their health issues also tested. Some other
processed by the ensemble hybrid algorithm and the results hybrid models can be developed and more real data sets
were compared. It is identified the performance evaluation with greater size can be collected and analyzed.
of the hybrid model (HML-EVC Model) performs best than
the individual models. References
[1] Taiwo Olapeju Olaleye, Olufunke Rebecca Vincen , “A Predictive
Model for Students’ Performance and Risk Level Indicators Using
Machine Learning”, IEEE, 2020.
[2] Fatai Anifowose, “Hybrid Machine Learning Explained in
Nontechnical Terms”, Data science and Digital Engineering, 2020.
[3] Sahil Parab, Piyush Rathodi, Durgesh Patil and Vishwanth
Chikkareddi,” A Multilayer Hybrid Machine Learning Model for
Diabetes Detection -2020.
[4] Jason Brownlee,” Stacking Ensemble Machine Learning With
Python, 2020.
[5] Vinutha K, Yogisha H K, “Employability Prediction of Engineering
Fig. 3: Comparison of Results Graduates using Machine Learning Algorithms”, International
Journal of Recent Technology and Engineering (IJRTE) ISSN: 2277-
VII. Results and Discussion 3878, Volume-8 Issue-5, January 2020.
[6] Tek Bist Bithari, Sharan Thapa, and Hari K.C.,” Predicting Academic
From the data set collected it is analyzed and the risk Performance of Engineering students using Ensemble method.
of the students is predicted in various aspects. [7] Hugo, Linsy S., M.S., “Analytics A Comparison of Machine
• Device Used: 61.7% of the students use smart Learning Models Predicting Student Employment”, Individual
phones for their education. Usage of smart phones Interdisciplinary Program, Master of Science in Higher Education
and Business Ohio University, December 2018.
can cause diabetes, obesity, heart disease, and [8] A. Salini, U.Jeyapriya,” A Majority Vote Based Ensemble Classifier
other health problems. for Predicting Students Academic Performance, International Journal
• 42% of students will interact with their parents of Pure and Applied Mathematics Volume 118 No. 24 2018.
less than an hour. This creates communication [9] Neeraj Khadilkar and Deepali Joshi. “Predictive Model on
Employability of Applicants and Job Hopping using Machine
gap between the parents and the children and the Learning”. International Journal of Computer Applications
children are not sharing their emotions to their 171(1):37-41, August 2017.
parents. [10] [2] Kajaree Das1, Rabi Narayan Behera, “A Survey on Machine
• 35.3% of the students have health issues like head Learning: Concept, Algorithms and Applications”, International
Journal of Innovative Research in Computer and Communication
pain, eye dryness and 24.7% are using spectacles. Engineering, Vol. 5, Issue 2, February 2017.
VIII. Conclusion and Future Work [11] Nishchol Mishra1, Dr.Sanjay Silakari,” Predictive Analytics:
A Survey, Trends, Applications, Oppurtunities & Challenges,
This research paper indicates the HML-EVC machine (IJCSIT) International Journal of Computer Science and Information
learning model is implemented with the student data to Technologies, Vol. 3 (3), 2012.

Copyright©IEEE–2022 ISBN: 978-1-6654-8734-4 | 691

Authorized licensed use limited to: Sri Sai Ram Engineering College. Downloaded on November 19,2024 at 08:40:02 UTC from IEEE Xplore. Restrictions apply.