![The block diagram for the proposed system III. RESULTS AND DISCUSSION The validation of SVM and RF in discriminating the SAA has produced good results across various types of research. A research article [26] presented some key validation tests of the ML systems in SAA classification [26]. The performance of SVM and RF for the identification of the SAA is validated based on prediction accuracy.](https://www.researchgate.net/profile/Keshav-Kaushik-2/publication/381319227/figure/fig1/AS:11431281251247351@1718162606823/The-block-diagram-for-the-proposed-system-III-RESULTS-AND-DISCUSSION-The-validation-of_Q320.jpg)
Content uploaded by Keshav Kaushik
Author content
All content in this area was uploaded by Keshav Kaushik on Jun 12, 2024
Content may be subject to copyright.
2024 2nd International Conference on Advancement in Computation & Computer Technologies (InCACCT)
979-8-3503-7131-4/24/$31.00 ©2024 IEEE
Evaluation of Random Forest and Support Vector
Machine Models in Educational Data Mining
Tsehay Admassu Assegie
School of Electronics Engineering,
Kyungpook National University,
Daegu, Republic of Korea
tsehayadmassu2006@gmail.com
Ayodeji Olalekan Salau
Department of Electrical and Computer
Engineering,
Afe Babalola University,
Ado-Ekiti, Nigeria
ayodejisalau98@gmail.com
Gunjan Chhabra
Department of CSE, Graphic Era Hill
University,
Dehradun, Uttarakhand, India,
chhgunjan@gmail.com
Keshav Kaushik
School of Computer Science, University of
Petroleum and Energy Studies,
Dehradun, Uttarakhand, India
officialkeshavkaushik@gmail.com
Sepiribo Lucky Braide
Department of Electrical and Electronics
Engineering, Rivers State University,
Port Harcourt 5080, Nigeria
braidesepiribo@yahoo.com
Abstract—The computer science field has witnessed the
popularity of machine learning (ML) in discriminating low
achieving and high-achieving students. However, various ML
methods have different performances in predicting student
performance. Therefore, the investigative analysis of their
effectiveness in the discrimination of student based on their
academic achievement would have been the major research
concern these days. This study investigates the performance of
the random forest (RF) and support vector machine (SVM)
against their power in academic performance prediction of a
student grade score (SGS). The analysis is performed based on
the classification capability of the two algorithms using the
Portuguese SGS dataset. Furthermore, the study also focused on
the analysis of the impact of sigmoid and radial basis functions
on the capability of the SVM for classifying SGS. We also
presented a comparison among the various ML methods namely
RF, and SVM, in identifying the student performance based on
the SGS. Various demographic information (age, sex) and
student assessment results (assignment, mid-term exam, and
quiz) were used as the features in training. The result revealed
that RF and SVM classifiers have the power to predict student
performance. The SVM scored more accuracy than the RF. We
obtained high accuracy (75.72% ) using the linear kernel. The
result implied that SGS can be predicted by using previous
assessment results with the proposed SVM classifier.
Keywords—data mining, quality of education, education
student performance, classification
I. INTRODUCTION
The past few years have experienced extensive research in
the analysis and classification of student academic
achievement (SAA). The applicability of the ML system has
become significant in predicting SAA and providing early
information for low-achieving students [1]. Additionally, the
availability of a large volume of data in the educational
landscape has paved the way for the ML capability to
disseminate the low and high achievers possibly aiding the
analysis and extraction of knowledge about the factors
influencing SAA.
Hence, because of their highly accurate discriminative
capability, the use of ML systems has become prominent in
the classification of high and love achieving students.
Moreover, these systems also aid in the investigation of the
influential factors on SAA. Thus, the evaluation of various
ML algorithms become one of the most important research
topics in machine learning. In [2], the researchers investigated
the effectiveness of SVM, K-Neighbors, Naïve Bayes (NB)
Artificial Neural Network (ANN), and decision tree (DT) for
the classification of the SAA. The study highlighted that the
ANN model has higher performance compared to the other
model.
The application of ML gained much research attention in
improving the SAA at the higher institution. Authors in [3]
applied machine-learning methods to predict student dropout
at a higher education institution. The study analyzed the
accuracy of RF, SVM, DT, and ANN on student dropout
prediction. The result appears to prove that the RF classifier
outperforms the RF, SVM, and DT. The discriminative power
of the low and high achievers using the RF is 70.98% accuracy
on the test data used in the research.
Lau et al. [4] employed ANN to develop a discriminatory
system of high and low-achieving students. The study
showcased that the ANN has been one of the dominant SAA
assessment techniques. The discrimination of low and high-
performing students with ANN helps the teachers to
differentiate low achievers before their failure by delivering
compositing and other supportive sessions to aid them in
improving their academic achievements. The experiment
revealed that ANN has an accuracy of 84.8% on SAA
classification.
Similarly, a research study [5] investigated a literature
survey on the application of ML in improving the
discrimination of high and low-achieving students by
analyzing their score quality. The researchers also highlighted
in their findings that ML has a wide range of applications in
the educational sector for the discrimination of the SAA [6].
The study highlighted that the ensemble learning methods are
one of the most commonly applied machine learning methods
for predicting student's academic performance.
SVM has also been used for developing an ML system to
identify the SAA by analyzing student performance data. It is
used for the classification of student grade scores as pass or
fail based on certain previous records [7]. I.K. Nti et al. [8]
supported the claim that ML systems have the power to
discriminate students based on performance by comparing
SVM with linear regression. The paper suggested that the
result showed that SVM had a lower mean square error in
classifying those students likely to drop out.
Another study in [9] implemented a DT-based predictive
model for SAA. The study [10] employed students’ previous
131
2024 2nd International Conference on Advancement in Computation & Computer Technologies (InCACCT) | 979-8-3503-7131-4/24/$31.00 ©2024 IEEE | DOI: 10.1109/InCACCT61598.2024.10551110
Authorized licensed use limited to: University of Petroleum & Energy Studies. Downloaded on June 12,2024 at 03:14:04 UTC from IEEE Xplore. Restrictions apply.
grades to predict the grade score of the student in future
subjects. The experimental result demonstrates the
implemented DT model achieves 63.63% accuracy on SAA
prediction. The study suggested that the implemented model
is helpful to the administration in evaluating and assessing the
results of students in their decision-making.
Higher education institutions have their standards for
evaluating student success [11]. However, they lack proper
procedures for handling student data and analyzing the
academic achievement of their student [12]. Thus, machine-
learning approaches have significance in implementing
appropriate procedures for predicting student academic
achievement in higher education institutions [13]. Research
articles [14] developed DT and RF-based models for
predicting the SAA. The study compared the performance of
DT with RF [15]. The result of the comparison with accuracy
as a performance metric demonstrates that the RF model
achieves a higher accuracy of 69.9% on student academic
performance (SAA) prediction [16].
The comparative study of different machine learning
approaches for predicting SAA conducted in different studies
[17] suggests that supervised learning methods have become
significant in discovering correlations among different
attributes of the student. The application of machine learning
approaches to student data assists in enhancing the quality of
academic performance of higher institution students.
Associative classification has become one of the significant
tools for predicting SAA [18]. A comparative study [19] on
different supervised learning methods such as DT, RF, NB,
and deep learning shows promising results. The RF model
gives an accuracy of 75.52% in predicting the SAA [20].
The literature survey in [21] shows that machine-learning
approaches are widely applicable to the educational sector for
analyzing educational data to improve education quality.
Thus, this study aims to investigate the RF and SVM
models. Overall, the objectives of this study are discussed as
follows: (1) To study the performance of RF, and SVM for
predicting the performance of student success. (2) To study
the effect of the SVM regularization parameters on the
performance of SVM. To study the effect of the depth of the
tree on its discriminative power of discrimination the SAA
with the RF and SVM. The organization of the research is as
follows: we described the method and the source of data in
section 2 while section 3 focuses on the comparative result
analysis, and section 4 covers the conclusion. The research
focused on the investigation of the discriminative power of RF
and SVM SAA, using various performance indicators, such as
accuracy, and the parameters of SVM which include radial
basis, polynomial, and sigmoid functions.
II. METHODOLOGY
The study procedures followed in conducting this analysis
of the RF and SVM in the discrimination of the student against
their SGS involved the following steps as suggested by a study
[22]. The collection of the dataset is conducted in the first step.
Firstly, we gathered the SAA dataset, including their
demographic data, grade scores, attendance records, and
related variables that have an impact on the SGS as previously
used in [23]. In the second step, we conducted preprocessing
of the collected dataset by cleaning missing values, removing
redundant data samples, and removing outliers, and
categorical variables. The dataset is collected from Portuguese
SAA obtained from the Kaggle data repository employed by
the previous study [24]. Thirdly, the dataset has split the
dataset into training and testing to train and validate the SVM
and RF. Then in the fourth step, the RF, and SVM are trained
on the training set using varying hyperparameters to obtain
good discriminative power of the employed ML methods.
Various validation methods such as accuracy, and confusion
matrix are used in analyzing the effectiveness of the SVM and
RF on the test set, these measures help validate the predictive
power of the ML systems [25].
After obtaining the SAA the sex, gender, parent's
qualification, economic, and academic attributes are collected
from the Kaggle repository. To implement the selected ML
methods for discriminating the students with their SGS, SVM,
and RF we used Python 3.8 Programming Language using
Intel(R) Core (TM) i7-8550U CPU @ 1.80GHz 2.00 GHz
with 8GB RAM. We have removed redundant and missing
values from the collected dataset before training the SVM, and
RF. Additionally, we have label-encoded the categorical
features to feed the input data to the RF, and SVM. Figure 1
indicates the procedures we followed in implementing and
testing the RF and SVM-based predictors for SAA. The steps
involved in the process of conducting this study involved data
collection to validation as observed in Figure 1. Finally, the
model is validated against its prediction accuracy of SAA.
Fig. 1. The block diagram for the proposed system
III. RESULTS AND DISCUSSION
The validation of SVM and RF in discriminating the SAA
has produced good results across various types of research. A
research article [26] presented some key validation tests of the
ML systems in SAA classification [26]. The performance of
SVM and RF for the identification of the SAA is validated
based on prediction accuracy.
The comparative investigation of the RF and the SVM
(with sigmoid) has shown that the SVM classifier has better
discriminative power than the RF classifier in identifying the
SAA based on grade score. moreover, the results also
indicated that the SVM discriminative ability varies with the
variation of the parameters such as sigmoid, polynomial, and
radial basis functions. The higher discriminative power is
achieved by training the SVM with the sigmoid function as
compared with the other parameters. The performance of the
RF and SVM is presented in sections 3.1 and 3.2 respectively.
A. The Performance of the RF
RF is a popular machine learning algorithm that is widely
used in Educational Data Mining (EDM) due to its ability to
handle complex relationships in data, handle high-
dimensional feature spaces, and provide robust prediction [27].
132
Authorized licensed use limited to: University of Petroleum & Energy Studies. Downloaded on June 12,2024 at 03:14:04 UTC from IEEE Xplore. Restrictions apply.
It typically performs well in predicting student outcomes, such
as academic performance, dropout risk, or course completion.
Its ensemble nature, combining multiple decision trees, helps
reduce overfitting and improve prediction accuracy compared
to individual decision trees [28]. While RF is considered a
black-box model to some extent, it still offers some level of
interpretability through feature importance analysis [29].
Educators and researchers can gain insights into which
variables are most influential in predicting student outcomes
[30]. It can handle large datasets efficiently and is
parallelizable, making it suitable for processing vast amounts
of educational data. It can also handle imbalanced class
distributions, which are common in educational datasets [30].
Fig.2 demonstrates the performance of the proposed RF in
predicting the SAA. Fig. 2 indicates the random forest
achieves the highest accuracy of 84.02%, a minimum
accuracy of 59.72%, and an average accuracy of 71.84% in
predicting the performance of SAA. Thus, the model is
effective in predicting students learning outcomes even
though the model has scope for improvement, as an accuracy
score of 84.02% does not accurately predict the learning
outcomes of a student.
Fig. 2. The performance of the RF
B. The Performance of the SVM
The performance of the SVM model is analyzed on
different parameters. The SVM model achieves different
accuracies for different parameters such as the sigmoid,
polynomial, and radial basis function (RBF). The sigmoid
SVM model achieves a higher accuracy of 83.33% as
compared to the polynomial and RBF parameters. Figure 3
demonstrates the maximum, minimum, and average
accuracies of the SVM model using the sigmoid, polynomial,
and RBF for predicting SAA.
Fig. 3. The performance of SVM
C. Comparison of the RF and SVM
Table I indicates the accuracy achieved by the RF and
SVM models in predicting student academic performance. As
indicated in Table I, the SVM model achieves higher average
accuracy. The RF model achieves the highest accuracy but the
average accuracy of the RF model is lower than the SVM
model for predicting the learning outcomes of students.
Overall, the SVM model achieved 75.72% while the RF model
achieved 71.81% accuracy.
In terms of accuracy, the SVM has been shown to achieve
high prediction accuracy in tasks such as predicting student
academic performance, dropout, and learning styles. The
experiment appears to prove that SVM has a better capability
of handling noisy and highly correlated data. Handling of
noisy and higher correlation makes SVM a good choice for a
dataset with high dimensions such as the SAA. Furthermore,
the SVM has also shown high accuracy in discriminating
student outcomes, particularly with the kernel parameter
showing relationships between SAA and input features and
complex discrimination boundaries between good achievers
and those with lower scores. The linear kernel allowed SVM
to capture inherent patterns in the SAA data.
TABLE I. COMPARISON OF THE PERFORMANCE OF RF AND SVM.
Algorithm
Minimum accuracy
Maximum
accuracy
Average
accuracy
RF
59.72%
84.02%
71.81%
SVM
65.97%
84.72%
75.72%
The experimental result also showed that RF was found to
be robust against overfitting and making and it is hence a good
choice for predictive analysis tasks with high dimensionality
or noisy data [31]. Moreover, as an ensemble method, it also
combines various decision trees, which reduces variance and
improves the prediction of SAA.
Similarly, SVM is good for datasets with outliers as it can
handle high-dimensional data. The main maximization of the
objective in the SVM is to achieve good predictive accuracy
by finding the optimal decision boundary, leading to a good
discriminative capability. It is evident that both SVM and RF
have their strengths and weaknesses, and the selection of these
classifiers should be based on the type of the dataset being
used the task to be addressed, and the characteristics of the
dataset used in classification. However, further research is
needed to validate the potential of other ML classifiers for
SAA prediction. Fig. 4 indicates the comparison of the
performance of SVM and RF.
133
Authorized licensed use limited to: University of Petroleum & Energy Studies. Downloaded on June 12,2024 at 03:14:04 UTC from IEEE Xplore. Restrictions apply.
Fig. 4. The accuracy of RF and SVM for SAA
In conclusion, RF and SVM have been shown to have a
good predictive ability of the SAA, with each classifier
offering unique strengths in terms of prediction capability,
robustness, explainability, and computational complexity. The
choice between these algorithms would depend on the specific
characteristics of the dataset, the types of data of the value to
be predicted, and the priorities of the researcher or investigator.
By considering the trade-offs between these factors,
researchers can select a good-performing ML system for their
EDM application.
By following the procedures presented in section II, we
can effectively validate SVM, and RF in predicting the SAA
and assist in informed results on selecting the important ML
classifier for an SAA prediction. However, the methods
largely would depend on the specific characteristics of the
SAA as this study only validated their performance based on
SAA, the nature of the classification problem, and the trade-
offs between accuracy, transparency, and computational
complexity. Researchers and investigators in EDM can benefit
by considering the strengths and limitations of SVM, and RF
when choosing the better-performing classifier for their
specific needs.
IV. CONCLUSION
This paper proposed an RF and SVM-based model for the
identification of SAA. The study employed various
approaches of data pre-processing to improve the
discrimination ability of the proposed classifiers for
discriminating student-learning outcomes. Moreover, the
study investigated the parameters of SVM, and how the
performance of the SVM model is affected by the parameters
used during the training. The result demonstrates that the
linear SVM model performs better than the RF model
achieving an overall prediction accuracy of 75.72%. in
conclusion, the result of the experiment shows that supervised
learning methods such as RF, and SVM significantly assist in
improving the education quality at higher education
institutions providing higher predictive power. Overall, both
RF and SVM have shown promising results in predicting
students' CGPA. RF tends to perform well in handling noisy
data and large datasets, while SVM is effective in high-
dimensional spaces and non-linear data. However, the choice
between the two algorithms ultimately depends on the specific
characteristics of the dataset and the goals of the prediction
task.
The results show that both RF and SVM are effective in
predicting student performance (predicting student cumulative
grade point average), but RF performs RF as an ensemble
method that combines multiple decision trees, which reduces
the risk of overfitting and improves the generalization of the
model. SVM, on the other hand, is a powerful algorithm for
classification tasks, but it may not perform as well as Random
Forest when dealing with large datasets or noisy data. Overall,
the study concludes that the choice of algorithm should be
based on the specific problem being addressed and the
characteristics of the dataset. Further research is needed to
explore the potential of other machine-learning algorithms for
educational data mining.
REFERENCES
[1].
A. Almasri, E. Celebi, and R.S. Alkhawaldeh., “G. Nguyen et al.,
“Machine Learning and Deep Learning frameworks and libraries for
large-scale data mining: a survey,” Hindawi Scientific Programming,
vol. 2019, pp. 1–14, 2019, doi: https://doi.org/10.1155/2019/3610248.
[2].
Y.A. Alsariera, “Assessment and Evaluation of Different Machine
Learning Algorithms for Predicting Student Performance,” Hindawi
Computational Intelligence and Neuroscience, vol. 2022, no. 1, pp. 1–
11, 2022, doi: s https://doi.org/10.1155/2022/4151487.
[3].
K. Dake, and C. Buabeng-Andoh, “Using Machine Learning
Techniques to Predict Learner Drop-out Rate in Higher Educational
Institutions,” Hindawi Mobile Information Systems, vol. 2022, no. 1,
2022, doi: https://doi.org/10.1155/2022/2670562.
[4].
E.T. Lau, L. Sun, and Q. Yang, “Modelling, prediction and
classification of student academic performance using artificial neural
networks,” SN Computer Science, 2019, doi: |
https://doi.org/10.1007/s42452-019-0884-7.
[5].
P. Balaji et al., “Contributions of Machine Learning Models towards
Student Academic Performance Prediction: A Systematic Review,”
Applied Science 2021 doi: https://doi.org/10.3390/app112110007.
[6].
R. Hasan et al., “Student Academic Performance Prediction by Using
Decision Tree Algorithm,” IEEE, 2018.
[7].
M. Kamal et al., “Metaheuristics Method for Classification and
Prediction of Student Performance Using Machine Learning
Predictors,” Hindawi Mathematical Problems in Engineering, vol.
2022, pp. 1–5, 2022, doi: https://doi.org/10.1155/2022/2581951.
[8].
I.K. Nti et al., “An empirical assessment of different kernel functions
on the performance of support vector machines,” Bulletin of Electrical
Engineering and Informatics, vol. 10, no. 6, pp. 3403-3411, 2021, doi:
10.11591/eei. v10i6.3046.
[9].
M. Zaffar, and K.S. Savita, “A Study of Feature Selection Algorithms
for Predicting Students Academic Performance,” International Journal
of Advanced Computer Science and Applications, vol. 9, no. 5, pp.
541–549, 2019. [10] H. Gull et al., “Improving Learning Experience of
Students by Early Prediction of Student Performance using Machine
Learning,” IEEE, pp. 1-4, 2019.
[10].
F.J Kaunang et al., “Students’ Academic Performance Prediction using
Data Mining,” IEEE, 2019. [12] C.C Kiu et al., “Data Mining Analysis
on Student's Academic Performance through Exploration of Student's
Background and Social Activities,” IEEE, 2018, doi:
10.1109/ICACCAF.2018.8776809.
[11].
S. Biju, A.O. Salau, J.N. Eneh, V. E. Sochima, I. T. Ozue, “A Novel
Pre-Class Learning Content Approach for the Implementation of
Flipped Classrooms,” International Journal of Advanced Computer
Science and Applications (IJACSA), Vol. 11(7), pp. 131-136,
2020. DOI: 10.14569/IJACSA.2020.0110718
[12].
J. Sadowski, “Predicting Student Academic Performance in Computer
Science Courses: A Comparison of Neural Network Models,”
International Journal of Modern Education and Computer Science, vol.
1, no. 1, pp. 1–9, 2018, doi: 10.5815/ijmecs.2018.06.01
[13].
H. Karalar, C. Kapucu, and H. Gürüler, “Predicting students at risk of
academic failure using ensemble model during the pandemic in a
distance learning system,” International Journal of Education in Higher
Education, vol. 18, no. 63, pp. 1–18, 2021, doi
https://doi.org/10.1186/s41239-021-00300-y.
[14].
E. Alyahyan, and Dilek Düştegör, “Predicting academic success in
higher education: literature review and best practices,” International
Journal of Education in Higher Education, vol. 17, no. 3, pp. 1–21,
2020, doi: https://doi.org/10.1186/s41239-020- 0177-7.
[15].
D.S Maylawati et al., “Data science for digital culture improvement in
higher education using K-means clustering and text analytics,”
International Journal of Electrical and Computer Engineering., vol. 10,
no. 5, 2020, pp. 4569-4580, doi: 10.11591/ijece. v10i5.pp4569-4580.
134
Authorized licensed use limited to: University of Petroleum & Energy Studies. Downloaded on June 12,2024 at 03:14:04 UTC from IEEE Xplore. Restrictions apply.
[16].
A.S. Hashim, W.D. Awadh, and A.K. Hamoud, “Student Performance
Prediction Model based on Supervised Machine Learning Algorithms,”
Materials Science and Engineering, 2020, doi:10.1088/1757-
899X/928/3/032019.
[17].
L. Cagliero et al., “Predicting Student Academic Performance using
Associative Classification,” Applied Sciences., pp. 1–22, 2021, doi:
https://doi.org/10.3390/app11041420.
[18].
N.A. Yassein, R.G.Helali, and S.B. Mohomad, “Predicting Student
Academic Performance in KSA using Data Mining Techniques”
Journal of Information Technology & Software Engineering, vol. 7, no.
5, pp. 1–5, 2017, doi: 10.4172/2165-7866.1000213
[19].
L. Nanglae, “Determining patterns of student graduation using a bi-
level learning framework,” Bulletin of Electrical Engineering and
Informatics, vol. 10, no. 4, 2021, pp. 2201-2211, doi: 10.11591/eei.
v10i4.2502.
[20].
S.A. Alwarthan, N. Aslam, and I.U. Khan, "Predicting Student
Academic Performance at Higher Education Using Data Mining: A
Systematic Review," Hindawi Applied Computational Intelligence and
Soft Computing, 2022, doi: https://doi.org/10.1155/2022/8924028.
[21].
S. Leonelli and N. Tempini, “Predicting Student Performance to
Improve Academic Advising Using the Random Forest Algorithm,"
International Journal of Distance Education Technologies. vol. 20, no.
1, pp. 1-17, doi: https://orcid.org/0000-0001-8440-5889.
[22].
S. Huang, and J. Wei, "Student Performance Prediction in Mathematics
Course Based on the Random Forest and Simulated Annealing,"
Hindawi Scientific Programming, 2022, doi:
https://doi.org/10.1155/2022/9340434.
[23].
D.T. Ha et al., “An Empirical Study for Student Academic Performance
Prediction Using Machine Learning Techniques,” International Journal
of Computer Science and Information Security, vol. 18, no. 3, 2020.
[24].
A. Asselman et al., “Enhancing the prediction of student performance
based on the machine learning XGBoost algorithm,” Interactive
Learning Environments, vol. 31, no. 6, 2023.
[25].
A. Triayudi, and I Fitri, “Comparison of the feature selection algorithm
in educational data mining,” TELKOMNIKA Telecommunication,
Computing, Electronics and Control vol. 19, No. 6, December 2021,
pp. 1865~1871, DOI: 10.12928/TELKOMNIKA.v19i6.21594.
[26].
S.T. Ahmed, R. Al-Hamdani, and M.S. Croock, “Enhancement of
student performance prediction using modified K-nearest neighbor,”
TELKOMNIKA Telecommunication, Computing, Electronics and
Control Vol. 18, No. 4, August 2020, pp. 1777-1783, DOI:
10.12928/TELKOMNIKA.v18i4.13849.
[27].
M. Yağcı, “Educational data mining: prediction of students' academic
performance using machine learning algorithms,” Smart Learn.
Environ. 9, 11 (2022). https://doi.org/10.1186/s40561-022-00192-z
[28].
W. Xing, R. Guo, E. Petakovic & S. Goggins, “Participation-based
student final performance prediction model through interpretable
Genetic Programming: Integrating learning analytics, educational data
mining and theory. Computers in Human Behavior, 47, pp. 168–181,
2015.
[29].
P. Dabhade, R. Agarwal, K.P. Alameen, A.T. Fathima, R. Sridharan,
G. Gopakumar, “Educational data mining for predicting students’
academic performance using machine learning algorithms,” Materials
Today: Proceedings, 2021. doi: 10.1016/j.matpr.2021.05.646
[30].
P. Chaudhury, and H.K. Tripathy, “A novel academic performance
estimation model using two-stage feature selection,” Indonesian
Journal of Electrical Engineering and Computer Science, Vol. 19, No.
3, September 2020, pp. 1610-619, DOI: 10.11591/ijeecs. v19.i3. pp
1610-1619.
[31].
S. Mohamed, and A. Ezzati, “A data mining process using
classification techniques for employability prediction, and Future
Opportunities,” Indonesian Journal of Electrical Engineering and
Computer Science, vol. 14, no. 2, May 2019, pp. 1025-1029, DOI:
10.11591/ijeecs. v14.i2. pp1025-1029.
135
Authorized licensed use limited to: University of Petroleum & Energy Studies. Downloaded on June 12,2024 at 03:14:04 UTC from IEEE Xplore. Restrictions apply.
