Machine Learning Applications in Student Academic Performance
Prediction: A Study in UDNR, Myanmar
Soe Moe Lwin1 , Myo Thant2 , Myat Nyein Moe3 , Zin Mar Oo4
1,2,3
Department of Information Technology, Defence Services Technological Academy, Pyin Oo Lwin, Myanmar
4
Graduate School of Economics, Ritsumeikan University, 1-1-1 Noji-Higashi, Kusatsu, Shiga 525-8577 Japan
1
smoelwin@iuj.ac.jp, 2 myothant82@gmail.com, 3 myatnyeinmoe82@gmail.com, 4 zinmaroo@iuj.ac.jp
A BSTRACT
In this study, we rigorously assess the predic-
tive power of five prominent machine learning al-
gorithms—Artificial Neural Network (ANN), Support
Vector Regression (SVR), Random Forest Regression
(RF), Gradient Boosting Regressor (GBR), and Ex-
treme Gradient Boosting (XGBoost)—in forecasting
student academic performance. Utilizing survey data
from the University for the Development of the National
Races of the Union (UDNR) in Myanmar, we undertake
a comparative analysis to delineate the most efficacious
model. This is followed by a detailed exploration of
feature importance based on the superior model identi-
fied. Our results unequivocally indicate that SVR out-
performs the other contenders in predictive accuracy, al-
though the remaining models also demonstrate usabil-
ity. Subsequent Permutation Feature Importance (PFI)
analysis reveals that lagged GPA is the most significant
predictor, followed in sequence by total marks obtained
in grade 11, gender, weekly study hours, and ethnicity.
The insights derived from this research highlight the
significant implications of machine learning method-
ologies in enhancing the accuracy of student academic
performance predictions, thereby potentially fostering
more personalized and effective educational strategies.
Keywords: academic performance, machine learn-
ing, feature importance, survey data analysis, Myanmar
I. INTRODUCTION
Students’ academic achievements in colleges and
universities have significant effects on their life out-
comes, including employment, income, and social sta-
tus, as well as success in the workforce and in con-
tributing to building a developed community and na-
tion (Adelman, 1999; Csikszentmihalyi and Schneider,
2000; Lee and Frank, 1990), as cited in Riegle-Crumb
(2006). Given the value of higher education for per-
sonal and socio-economic growth, many researchers
have emphasized the need to understand the factors that
can affect academic performance. Thus, predicting aca-
demic performance has become crucial for enhancing
educational outcomes and providing valuable insights
to educational institutions.
1
Factors that influence academic performance encom-
pass internal factors, such as a student’s innate ability
and self-motivation, and external factors, including res-
idential area, ethnicity, and gender (Kudari, 2016). Ad-
ditionally, the grade point average (GPA) plays a pivotal
role in both assessing a student’s past academic achieve-
ments and evaluating their future potential. Such as-
sessments are critical for various purposes, including
college admission, graduate program acceptance, schol-
arship awards, and entry into training programs and the
workforce, as discussed by Volwerk and Tindal (2012).
While other measures and outcomes may also be con-
sidered, GPA often takes precedence due to its simplic-
ity, numerical nature, and ease of comparison. Rec-
ognizing and accounting for these factors is essential
for accurately estimating a student’s academic perfor-
mance.
In this context, our study aims to use machine learn-
ing to predict student academic performance. We used
data from the University for the Development of the
National Races of the Union (UDNR) in Myanmar, en-
compassing information from 735 students. We tested
several machine learning algorithms, including Artifi-
cial Neural Networks and Support Vector Machines.
Through rigorous methodologies that include data sam-
pling, hyperparameter tuning, and model evaluation, we
aim to determine which techniques are most effective
for predicting student performance.
Our primary objectives in this study are to: (1)
demonstrate the predictive capabilities of modern ma-
chine learning models in forecasting student perfor-
mance, and (2) elucidate the various factors that shape
student outcomes. Preliminary findings underscore the
potential of machine learning in accurately forecasting
academic performance, with certain algorithms exhibit-
ing remarkable accuracy and reliability. Importantly,
our study highlights the significant roles of past aca-
demic performance, study hours, gender, and ethnicity
in academic achievement predictions. The subsequent
sections delve into the methodology, data, and results
that underpin these insights.
 II. LITERATURE REVIEW
In recent years, there has been growing interest in
utilizing machine learning and data mining techniques
to predict student academic performance. These stud-
ies, diverse in nature, employ a range of methodologies
and focus on various global contexts. Notably, current
empirical studies have yet to explore the application of
machine learning techniques for academic prediction
within the context of Myanmar.
Albreiki et al. (2021) conducted a comprehensive re-
view of key studies between 2009 and 2021, shedding
light on the expanding role of machine learning in ed-
ucational areas. The results affirm the transformative
potential of machine learning. More specifically, edu-
cators who leverage machine learning can gain a clearer
understanding of student progress, enabling early inter-
ventions for students facing challenges.
A study from the Obafemi Awolowo University in
Nigeria compared two neural network models: the
Multilayer Perceptron and the Generalized Regression
Neural Network. The objective was to determine
the most effective model for predicting students’ aca-
demic performance using just one performance met-
ric—academic results. The findings highlighted the
promising utility of machine learning methodologies
for educators in forecasting student performance Iyanda
et al. (2018).
Similarly, Yakubu and Abubakar (2022) used ma-
chine learning on data from a Nigerian university to
predict student performance, utilizing early indicators
such as age, gender, and previous academic scores. The
study highlighted the influential roles of gender, high
school examination scores, and region in determining
academic achievement. It suggests that these tools can
be crucial for higher education institutions when allo-
cating resources and devising intervention strategies.
Chen and Ding (2023) emphasizes the significance of
predicting academic performance for policymaking. In
an attempt to predict academic performance in Penn-
sylvania, various machine learning models were em-
ployed: decision tree (48% accuracy), random forest
(54% accuracy), logistic regression (50% accuracy),
support vector machine (51% accuracy), and neural net-
work (60% accuracy). Among them, the neural network
demonstrated the highest accuracy, highlighting its po-
tential as a valuable tool in shaping educational policies,
such as funding allocation and teacher selection.
In summary, the current literature underscores the
growing enthusiasm and significant advancements in
harnessing machine learning techniques within the ed-
ucation domain, especially for predictive analytics con-
cerning student performance. However, most existing
studies focus on specific institutions or distinct geo-
graphic areas. This reveals a gap in the research, sug-
gesting opportunities for broader analyses that cover
a range of educational settings, including places like
Myanmar.
2
III. DATA
The dataset used for this research was acquired be-
tween July 12 and July 25, 2018, from a tertiary ed-
ucational institution— the University for the Develop-
ment of the National Races of the Union (UDNR) —lo-
cated in the north-western region of Sagaing, Myan-
mar. Our sample comprises 735 students from the third,
fourth, and fifth academic years. Founded in 1964
as the Academy for Development of National Groups
(ADNG), the primary objective of UDNR is to nur-
ture an educational workforce dedicated to the socio-
economic development of national races, promoting
unity and prioritizing the progress of border areas, par-
ticularly in education and other social sectors.
The unique composition of the university, with its
blend of ethnic majority and minority groups, presents
an intriguing backdrop for our research. This diversity
stems from UDNR’s policy of selecting students based
on household income and ethnic affiliations. The insti-
tution ensures that a significant segment of its student
body comes from low-income families living in border
areas. Furthermore, deliberate efforts are in place to
achieve a balance across the eight major ethnic groups
in each class. Students are randomly assigned to one of
six classes (A to F), guaranteeing that the ethnic distri-
bution is independent of other factors that might affect
their GPAs.
Our research highlights the underrepresented ethnic-
ities of Kachin, Kayah, Kayin, Chin, Mon, Bamar,
Rakhine, and Shan. Information about students’ back-
grounds was collected using a structured questionnaire,
administered to third (grade 14), fourth (grade 15), and
fifth (grade 16) year students enrolled between 2015
and 2017. Of the respondents, 191 were male and
544 were female. Their standardized test scores, repre-
sented as GPA and used as the outcome variable in our
analysis, were sourced from the administrative records
of the respective academic years.
Our aim was to identify the factors affecting the aca-
demic performance of these advanced students. The
reason for focusing on these cohorts lies in the extensive
history of their academic data, especially their GPAs
from previous years. A ”lag” approach was adopted,
using prior GPAs as predictors for subsequent academic
years. Consequently, first and second-year students
were not considered. First-year students did not have
a GPA history, and the data for second-year students
was not extensive enough for a thorough prediction.
This approach resulted in one observation from each
third-year student, two from each fourth-year student,
and three from every fifth-year student, accumulating a
dataset of 1333 observations for the 735 students.
The features considered in our analysis include GPA
history, total marks of grade 11, weekly study hours,
ethnicity, gender, religious affiliation, residential back-
ground, and the location of the Basic Education High
School (BEHS). Each feature provides distinct insights.
For example, GP A lag and T otalmarksGrade11 re- crafted to ensure the developed models are robust, ac-
flect the impact of past academic achievements on the curate, and reliable in their predictions.
current GPA, while Studyhrperweek might indicate stu-
dents’ dedication. Variables related to ethnicity, reli- 1. Data Sampling: In this initial stage, the stu-
gion, and gender offer socio-demographic perspectives, dent academic performance data, collected from
whereas residential and BEHS location variables pro- the University for the Development of the Na-
vide contextual insights. Before starting the training tional Races of the Union (UDNR) in Myanmar,
process for our algorithms, we transformed our fea- are segmented into five subsets using the K-fold
ture data to suitable formats. For numerical attributes, cross-validation technique, with a specific choice
we applied the z-score normalization technique. This of K=5. This technique ensures that every data
method recalibrates the data by subtracting the mean of point is used for validation exactly once while the
each feature and then dividing the result by its standard remaining data points form the training set.
deviation. For categorical features, we employed var- 2. Hyperparameter Tuning: In the second stage,
ious encoding strategies, such as binary, ordinal, and Grid Search with K-fold cross-validation is em-
one-hot encoding, depending on the structure of the ployed to find the optimal hyperparameters for
data and the targeted algorithmic outcome. each machine learning model. This exhaustive
It’s important to note that our dataset is comprehen- searching method considers all possible combina-
sive, with no missing values across all observations and tions of the hyperparameters to find the combina-
features. This completeness allowed us to conduct our tion that minimizes the error, thus improving the
analyses without resorting to data imputation, ensuring prediction performance of the model. Table 2 de-
the credibility of our findings. tails the hyperparameters employed for each algo-
rithm within the grid search, along with the se-
Table 1: Summary Statistics lected parameters used in the model training pro-
Variable Obs. Mean Std Min Max cess.
GPA 1,333 4.3 0.6 3.0 5.0
GPA lag 1,333 -0.1 1 -2.0 1.2
TotalmarksGrade11 1,333 -0.1 1 -2.5 2.5 Table 2: Hyper-parameters
Studyhrperweek 1,333 1.8 1.7 0.0 6.0 Grid Choice
ownethpct 1,333 0.0 1 -1.7 2.8
search model
Gender
male 1,333 0.3 0.4 0 1 ANN
Religion - No. of nodes in 9, 18, 27 9
Buddhism 1,333 0.8 0.4 0 1 1st hidden layers
Christianity 1,333 0.2 0.4 0 1
Other 1,333 0.0 0.1 0 1
- No. of nodes in 4, 9, 18 4,9
Ethnicity 2nd hidden layers
Bamar 1,333 0.2 0.4 0 1 SVR
Chin 1,333 0.1 0.3 0 1 - C value 1, 5, 10 1
Kachin 1,333 0.1 0.2 0 1
Kayah 1,333 0.04 0.2 0 1 - Gamma value scale, auto auto
Kayin 1,333 0.1 0.3 0 1 RF
Mon 1,333 0.02 0.2 0 1 - Maximum depth None, 5, 10 5
Rakhine 1,333 0.1 0.3 0 1 - No. of estimators 100, 200, 300 100, 200, 300
Shan 1,333 0.3 0.5 0 1
Residential] GBR
rural 1,333 0.6 0.5 0 1 - Learning rate 0.01, 0.05, 0.1 0.01
suburban 1,333 0.3 0.5 0 1 - No. of estimators 300, 500, 1000 300, 500
urban 1,333 0.1 0.3 0 1
XGBoost
LocationofBEHS
centeroftown 1,333 0.1 0.3 0 1 - Maximum depth 6, 8, 10 6
isolatedarea 1,333 0.1 0.3 0 1 - No. of estimators 500, 1000, 1500 500, 1000
outskirtoftown 1,333 0.4 0.5 0 1
rural 1,333 0.3 0.5 0 1 Note: For the grid search, two parameters for each al-
gorithm were selected for tuning, while suitable values
were assigned to other parameters not included in this
IV. METHODOLOGY table.

A. Examination Procedures 3. Model Training: Once the optimal hyperparam-

The examination procedure employed in this study
for predicting student academic performance using ma-
chine learning models comprises four distinct stages:
Data Sampling, Hyperparameter Tuning, Model Train-
ing, and Model Testing. These stages are meticulously
eters are identified, each machine learning model
is trained using the cross-validation technique.
Specifically, for each of the five subsets created in
the first stage, a model is trained using the four re-
maining subsets as the training data, and the held-
out subset as the validation data. This process is

3
 repeated five times, once for each subset, resulting
in five models for each algorithm.
4. Model Testing: Finally, the saved models are
tested using the testing data, which is the one
fold left out in each iteration. The performance
of the models is evaluated using three evaluation
metrics: Coefficient of Determination (R2 ), Root
Mean Square Error (RMSE), and Mean Absolute
Error (MAE). These metrics provide a comprehen-
sive understanding of the model’s predictive capa-
bilities and enable a thorough comparison of the
performance of the five algorithms.
Pn
(yi − yˆi )2
R2 = 1 − Pi=1
n 2
(1)
i=1 (yi − ȳ)
r Pn
i=1 (yi − yˆi )2
RM SE = (2)
n ing that embeds regularization techniques to de-
Pn
i=1 |yi − yˆi |
M AE = (3)
n
where yi represents the observed data, yˆi repre-
sents the predicted data, ȳ is the mean of the ob-
served data, n is the total number of observations.
B. Machine Learning Algorithms
In this subsection, we outline the suite of machine
learning algorithms employed to predict student aca-
demic performance. We selected five notable algo-
rithms: Artificial Neural Network (ANN), Support Vec-
tor Regression (SVR), Random Forest Regression (RF),
Gradient Boosting Regressor (GBR), and Extreme Gra-
dient Boosting (XGBoost).
1. Artificial Neural Network (ANN): ANNs are
mathematical models inspired by the functioning
of human brains (Abu-Naser et al., 2015). Intro-
duced in 1943, their adaptability has facilitated di-
verse applications over the years. Comprising in-
terconnected nodes organized in layers, each node
performs simple computations. Information tra-
verses these connections, resulting in predictions.
During training, by fine-tuning the weights and bi-
ases of these nodes, ANNs can capture intricate
patterns and relationships embedded in the data.
2. Support Vector Regression (SVR): In contrast to
conventional linear regression, SVR aims to iden-
tify a curve or hyperplane that optimally aligns
with the data, while accommodating a certain de-
gree of error, denoted by the margin (Cortes and
Vapnik, 1995). SVR adeptly maps data into a
higher-dimensional space, facilitating an optimal
distinction between data points.
4
3. Random Forest Regression (RF): RF, an en-
semble technique, constructs an array of decision
trees and amalgamates their predictions (Breiman,
2001a). By instilling randomness in tree formula-
tion and considering only a feature subset at each
bifurcation, RF encourages diversity among trees,
mitigating variance.
4. Gradient Boosting Regressor (GBR): GBR is a
potent ensemble learning algorithm that sequen-
tially crafts multiple weak learners, predominantly
decision trees (Friedman, 2001). Each subsequent
tree rectifies the errors of its antecedent, incremen-
tally enhancing the model’s precision. By integrat-
ing the forecasts of these individual trees, GBR
yields a robust predictive model that adeptly dis-
cerns intricate variable interrelations.
5. Extreme Gradient Boosting (XGBoost): XG-
Boost is an evolved variant of gradient boost-
ter overfitting and bolster generalization (Chen and
Guestrin, 2016). It refines the model through
gradient-based methodologies and approximate
tree learning, culminating in an exceptionally ef-
ficient and precise predictor.
C. Feature Importance
Permutation Feature Importance (PFI) is a method-
ology used to evaluate the importance of features (i.e.,
input variables) in a machine learning model (Breiman,
2001b). In this approach, the value of a single feature
in the dataset is randomly shuffled, and the consequent
impact on the model’s performance is measured. By
scrambling the feature values, we break the relationship
between the feature and the target variable, which can
provide insight into how much the model’s predictions
depend on the feature. If shuffling a feature’s values
leads to a significant decrease in the model’s perfor-
mance, it indicates that the model relies heavily on that
feature for making predictions. Conversely, if shuffling
a feature’s values has a minimal impact on the model’s
performance, it suggests that the feature is not impor-
tant. The process is repeated for each feature in the
dataset, providing a ranking of the features based on
their importance. PFI provides a practical way to un-
derstand the contribution of each feature to the model’s
predictive power, which is particularly useful when in-
terpreting complex models and improving their perfor-
mance.
V. RESULTS
A. Prediction Performance
Based on Table 3 detailing the prediction perfor-
mance of five machine learning algorithms in forecast-
ing academic performance (measured via GPA). For the
training data, XGBoost exhibited the most promising B. Feature Importance
performance with the lowest MAE of 0.16 and RMSE
of 0.22. Additionally, its R2 value of 0.88 was notably Feature importance illustrates the extent to which a
higher than the other models, suggesting a superior fit to particular variable aids in the predictions made by the
the training data. Regarding the testing data, the models model, defining the relative utility of every feature in
generally showcased closely matched results. However, the model’s predictive process. In Figure 2, we can ob-
the Support Vector Regression (SVR) stands out with an serve two distinct PFI plots which delineate the vital
MAE of 0.22, RMSE of 0.30, and an R2 of 0.77, mak- elements in our academic performance analysis. The
ing it the best-performing model for the testing dataset. left panel of the diagram arranges the important features
Given SVR’s promising results, we will further utilize based on the change in RMSE loss before and after the
this model to conduct a feature importance analysis us- permutation of features. The right panel portrays the
ing the Permutation Feature Importance (PFI) method. feature importance through the respective RMSE fol-
In Figure 1, five plots visually contrast the actual versus lowing each feature permutation.
predicted GPA outcomes for the testing data, as derived
Feature importance (type=ratio) Feature importance (type=raw)
from each of the five machine learning algorithms. GPA_lag 2.703 GPA_lag 0.769
TotalmarksGrade11 1.075 TotalmarksGrade11 0.305
Gender_male 1.037 Gender_male 0.295
Studyhrperweek 1.034 Studyhrperweek 0.293
Ethnicity 1.024 Ethnicity 0.291
Residential 1.023 Residential 0.29
LocationofBEHS 1.017 LocationofBEHS 0.289
ownethpct 1.015 ownethpct 0.288
Religion 1.006 Religion 0.285
Full model 1.0 Full model 0.284
1.0 1.5 2.0 2.5 0.3 0.4 0.5 0.6 0.7 0.8
Root mean square error (RMSE) Root mean square error (RMSE)

Figure 2: Feature importance for GPA prediction (SVR

model)

In the left panel, the RMSE loss after accounting

for the first important feature, GP A lag, stands at
2.703, while it is 0.769 in the right panel. This in-
dicates that the RMSE of our model escalated from
0.284 to 0.769, experiencing a 2.7-fold increase follow-
ing the permutation of the GP A lag variable. Follow-
ing the GP A lag, the features T otalmarksGrade11,
Gender, Studyhrperweek, and Ethnicity rank as
the second to fifth most important factors in predicting
GPA, which is a measure of academic performance. It
is important to note that PFI only demonstrates the rela-
tive importance of a variable in model predictions with-
out delineating its positive or negative influence on the
projected academic performance.
Moreover, it is important to note that when we ana-
Figure 1: Performance comparison of 5 algorithms on lyzed other algorithms, we found a similar pattern —
testing data the top features identified as important in this model
were often highlighted in the other models as well.
While the exact rankings differed slightly between
models, they all agreed that these features are key fac-
Table 3: Prediction Performance tors in predicting GPA. This similarity across different
MAE RMSE R2
models reinforces the importance of these top features
Testing in understanding academic performance.
ANN 0.223 0.302 0.768
GBR 0.235 0.308 0.759 VI. CONCLUSIONS
RF 0.220 0.303 0.767
SVR 0.219 0.301 0.770 Through the comparative study of five machine learn-
XGBoost 0.234 0.315 0.749 ing models and the application of Permutation Feature
Training Importance (PFI) analysis, our research provides valu-
ANN 0.211 0.282 0.799 able insights into predicting student academic perfor-
GBR 0.208 0.267 0.819 mance at the University for the Development of the Na-
RF 0.190 0.258 0.831 tional Races of the Union (UDNR) in Myanmar, con-
SVR 0.199 0.279 0.802 tributing to data-driven decision making and enhancing
XGBoost 0.163 0.221 0.877 educational outcomes.
The study not only spotlighted the proficiency of the

5
SVR algorithm in anticipating student performance but
also showcased the admirable predictive performances
exhibited by the other four algorithms, which displayed
results nearly parallel to that of the SVR, denoting a
generally robust efficacy across the range of algorithms
deployed. This good fit against the testing data essen-
tially illustrates a promising avenue for leveraging ma-
chine learning techniques in educational settings, af-
fording a tool that can help in the accurate prediction of
student academic performance to aid educational strate-
gies.
For a more detailed understanding of the influenc-
ing factors behind student academic performance, fu-
ture research avenues could delve deeper by employ-
ing advanced interpretable machine learning methods.
Techniques such as Accumulated Local Effects (ALE)
(Apley and Zhu, 2020) and SHapley Additive exPlana-
tions (SHAP) (Lundberg and Lee, 2017) hold promise
in providing deeper insights, facilitating the uncovering
of intricate patterns and relationships. Leveraging these
advanced techniques could potentially foster a compre-
hensive understanding and guide the development of
strategies aimed at nurturing academic excellence.
ACKNOWLEDGEMENT
We extend our heartfelt gratitude to the University
for the Development of the National Races of the Union
(UDNR) for granting permission to conduct the survey
and collect essential data.
A special acknowledgement is also due to our co-
author, Zin Mar Oo. Beyond her contributions as an
author, her expertise and mentorship in machine learn-
ing techniques were indispensable, bridging the gap in
our initial understanding of the subject.
R EFERENCES
Abu-Naser, S. S., Zaqout, I. S., Abu Ghosh, M., Atallah,
R. R., and Alajrami, E. (2015). Predicting student
performance using artificial neural network: In the
faculty of engineering and information technology.
Albreiki, B., Zaki, N., and Alashwal, H. (2021). A
systematic literature review of student’ performance
prediction using machine learning techniques. Edu-
cation Sciences, 11(9):552.
Apley, D. W. and Zhu, J. (2020). Visualizing the effects
of predictor variables in black box supervised learn-
ing models. Journal of the Royal Statistical Soci-
ety: Series B (Statistical Methodology), 82(4):1059–
1086.
Breiman, L. (2001a). Random forests. Machine Learn-
ing, 45:5–32.
Breiman, L. (2001b). Statistical modeling: The two cul-
tures (with comments and a rejoinder by the author).
Statistical Science, 16(3):199–231.
Chen, S. and Ding, Y. (2023). A machine learning ap-
proach to predicting academic performance in penn-
sylvania’s schools. Social Sciences, 12(3):118.
Chen, T. and Guestrin, C. (2016). Xgboost: A scalable
tree boosting system. In Proceedings of the 22nd
acm sigkdd international conference on knowledge
discovery and data mining, pages 785–794.
Cortes, C. and Vapnik, V. (1995). Support-vector net-
works. Machine Learning, 20:273–297.
Friedman, J. H. (2001). Greedy function approxima-
tion: a gradient boosting machine. Annals of Statis-
tics, 29(5):1189–1232.
Iyanda, A. R., Ninan, O. D., Ajayi, A. O., and
Anyabolu, O. G. (2018). Predicting student academic
performance in computer science courses: A compar-
ison of neural network models. International Journal
of Modern Education & Computer Science, 10(6).
Kudari, J. M. (2016). Survey on the factors influ-
ences the students’ academic performance. Interna-
tional journal of Emerging Research in Management
& Technology, 5(6):30–36.
Lundberg, S. M. and Lee, S.-I. (2017). A unified ap-
proach to interpreting model predictions. Advances
in neural information processing systems, 30.
Riegle-Crumb, C. (2006). The path through math:
Course sequences and academic performance at the
intersection of race-ethnicity and gender. American
Journal of Education, 113(1):101–122.
Volwerk, J. J. and Tindal, G. (2012). Documenting stu-
dent performance: An alternative to the traditional
calculation of grade point averages. Journal of Col-
lege Admission, 216:16–23.
Yakubu, M. N. and Abubakar, A. M. (2022). Apply-
ing machine learning approach to predict students’
performance in higher educational institutions. Ky-
bernetes, 51(2):916–934.