264
www.amazoniainvestiga.info ISSN 2322- 6307
DOI: https://doi.org/10.34069/AI/2024.74.02.22
How to Cite:
Alhamad, I.A., & Singh, H.P. (2024). Predicting dropout at master level using educational data mining: A case of public health
students in Saudi Arabia. Amazonia Investiga, 13(74), 264-275. https://doi.org/10.34069/AI/2024.74.02.22
Predicting dropout at master level using educational data mining:
A case of public health students in Saudi Arabia
:
Received: January 7, 2024 Accepted: February 25, 2024
Written by:
Ibrahim Abdullah Alhamad1
https://orcid.org/0000-0001-7099-0335
Harman Preet Singh2
https://orcid.org/0000-0003-4297-0016
Abstract
Student dropout and its economic and social
consequences are significant issues in
developing countries. Students who drop out
experience reduced employment prospects and
encounter social stigma. While early dropout
prediction can assist in mitigating the
consequences, it remains a considerable
challenge. The present research employed a data
mining approach to predict dropout of public
health master-level students in Saudi Arabia, a
developing nation that has invested considerable
resources to promote higher education. The
research model focused on three fundamental
determinants of students’ dropout: individual,
institutional, and academic. The study analysis
on a dataset of 150 students revealed that all three
determinants predicted student dropout. The
results indicated that students with low academic
performance who received an academic warning
were likelier to drop out. Freshmen with poor
academic achievement were particularly at risk
of dropping out of college. Students between 31
and 36 years old who attended technical courses
as a subject specialization could also dropout.
The research contributes to the literature by
suggesting that universities should consider these
individual, institutional, and academic
determinants to develop their dropout prevention
strategies. This study has ramifications for
university administrators in developing nations,
such as Saudi Arabia, who can establish dropout
ةصخ
150
31
36
1
Department of Management and Information Systems, College of Business Administration, University of Ha'il, Kingdom of Saudi
Arabia. WoS Researcher ID: HII-6863-2022
2
Department of Management and Information Systems, College of Business Administration, University of Ha'il, Kingdom of Saudi
Arabia. WoS Researcher ID: B-7160-2012
Alhamad, I.A., Singh, H.P. / Volume 13 - Issue 74: 264-275 / February, 2024
Volume 13 - Issue 74
/ February 2024
265
http:// www.amazoniainvestiga.info ISSN 2322- 6307
prevention programs based on the determinants
revealed in this study.
Keywords: Dropout, educational data mining,
higher education institutions, public health,
Saudi Arabia, Saudi Vision 2030.
Introduction
Higher education is a vital constituent for the
progress of a country. Higher education is
particularly important for developing countries
like Saudi Arabia (Singh et al., 2011a). In 1932,
when the Kingdom was established, Saudi
Arabia had no university. King Saud University
was the inaugural university founded in Saudi
Arabia in 1957 (Al-Omar, 2023). Saudi Arabia
undertook an ambitious effort in the 1990s to
establish a vast array of educational facilities,
recognizing the importance of education to the
growth of the nation (Singh & Chand, 2012). In
Saudi Arabia, the proportion of education
spending to GDP climbed from 5.3% in 1985 to
7.3% in 2018 (Singh et al., 2022a).
This illustrates that Saudi Arabia invested
extensively in education to enhance human
capital and improve its human resources in
accordance with the ideals of the Vision 2030
government program (Singh & Alodaynan,
2023). Currently, Saudi Arabia has 67 public and
private universities and colleges that deliver
higher education to students (CGIJ, 2023).
Students' academic performance during
university education is a turning point in their
career (Singh et al., 2013). Students obtain
employment opportunities depending on their
academic performance at the university level
(Singh et al., 2022b). The improved performance
of students has a favorable effect on the
reputation of higher education institutions (HEIs)
(Singh & Alhamad, 2022a). However, student
dropout negatively influences the employment
prospects of the affected students and seriously
impacts a nation’s scarce resources (Alhamuddin
et al., 2023). Typically, student dropout
negatively affects the reputation of the institution
and leads to strategic and monetary loss for the
nation (Singh et al., 2011b). Further, dropout
reduces employment possibilities and leads to
societal stigma, resulting in negative social and
economic effects for the students (Ye et al.,
2022). This is especially critical in the case of
public health students, who, following
graduation, play a crucial role in educating the
population about diverse health issues and their
management. The dropout of public health
students diminishes the likelihood that
governments, particularly in developing nations
(Kumari & Singh, 2021) (like Saudi Arabia), will
produce sufficient health educators in the future.
The dropout of public health students from HEIs
is critical for Saudi Arabia, as it has invested
significant resources to manage public health
(Alam et al., 2022).
Prior research has thoroughly investigated the
issue of higher education student dropout.
However, most of the prior higher education
research has primarily focused on the dropout of
undergraduate university students (Baalmann,
2023; Abdulghani et al., 2023). Some studies
have examined master's student dropout
(Dlungwane & Voce, 2020; Rotem et al., 2020),
however, there is a shortage of empirical research
on the dropout of public health students,
especially in Saudi Arabia. The current study is
built on earlier studies but differs from the
existing ones in context (Saudi Arabia) and focus
(public health).
The current study also employs data mining
techniques (Singh & Alhamad, 2022b) to identify
the predictive factors affecting dropout of
students, which is unique in the Saudi context.
The rest of the paper is arranged as follows. The
second section deals with the study objectives.
The third section is dedicated to the literature
review on educational data mining and students’
dropout. The fourth section deals with the
methods employed in this study. The fifth section
is related to the results and discussion. The sixth
section presents the conclusion of the study. The
last section is dedicated to the limitations and
future research directions.
Objectives of study
This study aims to develop a predictive model
using educational data mining to predict
students’ dropout. The following are the
objectives of this study:
• To identify the important attributes that help
to predict the students’ dropout.
266
www.amazoniainvestiga.info ISSN 2322- 6307
• To identify an appropriate data mining
algorithm suitable for building a students’
dropout predictive model.
• To apply the selected algorithm to train, test,
and build the students’ dropout model.
• To suggest strategies to reduce the dropout
rate of students.
Literature review on educational data mining
Higher education is deemed essential for a
country's progress, particularly in developing
nations (Alhulail & Singh, 2023). Early dropout
prediction is crucial to mitigate its detrimental
impact on the standard of higher education. The
commitment of the faculty (Robert, 2023) and
technology (Singh et al., 2011c) can play a vital
role in the early identification of students at the
risk of dropping out (Alkhalil, 2021; Alazemi,
2023). Student dropout rates may negatively
impact individuals in developing nations such as
Saudi Arabia (Ibeaheem et al., 2018), as it
diminishes their job opportunities in nations with
constraints on economic resources (Singh &
Alwaqaa, 2023).
Educational data mining utilizes education-
related data to produce data mining results (Singh
et al., 2023; Alhamad & Singh, 2021). In line
with the prior research (Gentle et al., 2012; Singh
& Alhamad, 2021; Singh & Alhulail, 2023), we
employ predictive modelling in this study.
Consequently, we undertake a literature review
on using educational data mining (EDM) and
other empirical methods to examine critical
factors affecting student dropout.
Zhang et al. (2010) leveraged EDM to identify
key student dropout determinants and improve
retention. They conducted the study by collecting
institutional data from Thames Valley University
students in the United Kingdom (UK). They
employed classification algorithms like naïve
bayes, decision trees, and support vector machine
(SVM) to identify the key predictive factors. The
study results indicated that the students’ average
course marks were a key determinant of their
dropout. Further, the study informed that first-
year students are more likely to drop out.
Bharadwaj & Pal (2011) employed EDM to
conduct a study using institution-specific internal
factors to analyze and predict students’ retention
at VBS Purvanchal University, Jaunpur, India. In
their study, the classification method of data
mining was used to analyze the performance of
50 MCA students using the classification
approach of data mining. From the university
database, the study used data such as attendance,
class quizzes, seminars, and assignment grades to
predict the students' performance. The study
employed ID3, C4.5, and ADT decision tree
classifiers to predict student performance.
Applying the model to the student data of
incoming freshmen revealed that the ADT
algorithm generated a concise and precise
prediction list for student retention. In this study,
the decision tree classification technique was
tested and found to be highly accurate for
improving student retention.
Bharadwaj & Pal (2012) employed EDM to
conduct a study based on external factors to
predict the retention of students at RML Awadh
University, Faizabad, India. Data from 300
students from 5 different degree colleges
studying Bachelor of Computer Applications
(BCA) was collected for this study. A prediction
model for student performance enhancement was
developed utilizing the data mining classification
method. Bayesian classification model was
applied to 17 data attributes of the students.
Students' senior secondary exam marks, dwelling
location, teaching medium, specialized courses,
mother’s qualification, family income, and
family status were found to be highly connected
with their dropout.
Yadav et al. (2012) employed EDM to predict
university students’ retention using machine
learning classifiers. They utilized decision tree
algorithms like ID3, C4.5, and ADT to a dataset
of selected students from VBS Purvanchal
University in Jaunpur, India. The study informed
that students’ grades play a key role in their
continuation of studies. Pal (2012) conducted
another study to predict the dropout rate of
engineering students at VBS Purvanchal
University, Jaunpur, India. They employed
decision tree algorithms such as ID3, C4.5,
CART, and ADT to identify the key predictive
factors influencing the dropout of students. The
results of this study revealed that students’ high
school and secondary school grades, family
income, and mother’s education level impacted
their dropout.
Márquez-Vera et al. (2015) employed EDM to
predict early dropout among high school students
in Mexico. They proposed a methodology and an
ICRM2 algorithm to improve the accuracy of
early dropout prediction. They compared the
efficacy of their ICRM2 algorithm with
traditional data mining algorithms. The study
results informed that the developed ICRM2
algorithm could predict dropout within four to six
weeks of the start of the course and could be
Volume 13 - Issue 74
/ February 2024
267
http:// www.amazoniainvestiga.info ISSN 2322- 6307
employed to detect early dropout of students. The
study results also suggested that students’ grades,
absenteeism, and motivation level influenced
their dropout.
Casanova et al. (2018) utilized EDM to identify
the key determinants of first-year students’
attrition at a public university in Portugal. The
study employed IBM SPSS software version 24
to create a decision tree for predicting the key
determinants of students’ dropout. The study
results revealed that students’ academic
performance played a key role in determining
their dropout. Specifically, low-achieving first-
year students were at a higher risk of dropping
out. Further factors such as sex, course type, and
education level of the mother also affected the
dropout of first-year university students.
A study was carried out by Ahmed et al. (2018)
in Bahrain to examine the relationship between
academic support and student involvement. The
study focused on how academic elements
influence students' academic engagement and
performance. The study findings indicated that
increased academic engagement among students
contributes to their retention in higher education.
Therefore, students who receive academic
support and are academically engaged are less
likely to drop out.
Rodríguez-Muñiz et al. (2019) leveraged EDM to
identify the profiles of students at risk of
dropping out at the University of Oviedo, Spain.
They employed data mining algorithms such as
CART, C4.5, bayesnet, RF, and SVM to identify
the risk profiles. The study results informed that
students were at a higher risk of dropping out in
the first year of university. Specifically, the first-
year students with low academic performance
were at a higher risk of dropping out. The other
factors that influenced students’ dropout
included students full-time or part-time
education, age, and attendance.
Mubarak et al. (2020) utilized EDM to predict the
dropout of students studying online at the Open
University, UK. They employed data mining
algorithms such as SVM, random forest, and
decision trees to develop the early prediction
model. The study's results indicated that the
likelihood of a student dropping out of an online
course decreased as their participation in course
activities increased. The study also indicated that
the study models were 82 percent accurate in
predicting the chance of students dropping out
during the second week of the online course.
Singh & Alhulail (2022) conducted a study on
dropout of students in teacher-training colleges
in Ethiopia. The authors employed a four-step
logistic regression approach to predict the critical
determinants of students dropout in the context
of the least-developed country (Ethiopia). The
study found that academic variables, such as
academic performance and higher education
aspirations, have a crucial role in affecting
student dropout rates. The study found that
cultural influences, family education level, and
economic considerations did not significantly
affect dropout decisions in Ethiopian teacher
training colleges.
Hashim et al. (2024) did an empirical study to
evaluate the factors influencing student dropout
rates in Malaysia. They analyzed a database of
over 100,000 students to identify the dropout
determinants. The study indicated that academic
characteristics, students' gender, and family
financial level influence student dropout rates.
The literature review reveals that individual,
institutional, academic, and economic factors
predict dropout of students. The individual factor
comprises students' age, gender, and parents'
education. The institutional factor comprises the
student's field of study and the type of course.
The academic factor encompasses students'
marks, GPA, CGPA, and attendance. The
economic factor encompasses family income and
the availability of a job for the student. The
current study aims to evaluate the impact of these
factors on the dropout rate of public health
masters’ students in Saudi Arabia.
The literature review indicates a lack of empirical
studies in Saudi Arabia that have applied dropout
factors identified in the literature, including
individual, institutional, academic, and economic
aspects. Further, the majority of dropout studies
utilizing EDM have been undertaken outside
Saudi Arabia. In addition, there is a dearth of
dropout research employing EDM on students of
public health, particularly at the master's level.
Therefore, the data lies unused in Saudi
universities and is not mined to uncover hidden
knowledge regarding the causes of master's
students dropping out of public health programs.
It is crucial to analyze the factors influencing the
dropout of public health students in Saudi
Arabia, especially at the master's level, given
Saudi Arabia's substantial investment in higher
education. The insufficient research in this field
warranted the necessity of doing this study. This
study aims to fill this research gap.
268
www.amazoniainvestiga.info ISSN 2322- 6307
Methods
In this study, we adapted the six stages of the Cios
et al. (2000) model to predict the dropout of public
health students.
Stage 1 – Understand Problem Domain
In this stage, we reviewed the literature on EDM to
understand the important factors influencing
dropout of students. This stage clarified EDM goals
in line with Cios et al. (2007). This study considered
the individual, institutional, and academic
determinants of public health masters’ students at
the University of Ha’il (UoH), Saudi Arabia. We
did not consider the economic factors as
information such as income level was not retained
by the UoH, Saudi Arabia.
Stage 2 – Understand Data
At this stage, we collected academic data from the
UoH, Saudi Arabia. We obtained 161 records for
students pursuing a master's in public health. The
UoH kept the data of the public health students
related to individual, institutional, and academic
determinants. The factors corresponding to
individual determinants were student ID number,
national ID number, name, mobile number, email
address, gender, and date of birth. The students' ID
number, national ID number, name, mobile number,
and email address could have no influence on their
dropout, therefore we ignored them. The factor
corresponding to institutional determinant was the
subject specialty offered to public health masters’
students. The factors corresponding to academic
determinants were cumulative GPA (CumGPA) (0
to 4), academic warnings (0 to 2), number of credit
hours required (45), number of credit hours passed
(0 to 45), and number of credit hours remaining (0
to 45). The number of credit hours remaining can be
computed as a difference between the number of
credit hours required and passed; therefore, we
ignored the number of credit hours remaining
attribute.
Stage 3 – Prepare Data
In this stage, we evaluated the dataset for missing
values and outliers (Han et al., 2012). We found 11
missing values in the dataset; therefore, we ignored
those records. There were no outliers found in the
dataset. The selected dataset consisted of 150
records. The dataset values should be converted
from continuous to categorical for data mining tasks
(Liu et al. (2002). For this purpose, first, we derived
age, cumulative GPA, and percentage of credit
hours completed (CRPER) attributes. We derived
the age from the date of birth. The CRPER was
derived as a proportion of the number of credit
hours passed to the number of required credit hours.
We converted the cumulative GPA from a number
to a percentage, where a cumulative GPA of 4
denoted 100%. Next, we transformed the age,
specialty, cumulative GPA, and CRPER attributes
from continuous to categorical.
Table 1 portrays the criteria for data transformation.
Table 1.
Criteria for Data Transformation
Attribute(s)
Continuous Value(s)
Transformed Value(s)
Age
24 to 30
A1
31 to 36
A2
37 to 42
A3
Specialty
Electronic Health
S1
Health Informatics
S2
Health Service Management
S3
Hospital and Health Services
S4
Occupational Health
S5
Public Health
S6
Cumulative GPA (CumGPA)
>95%
A+
90% to 94.99%
A
85% to 89.99%
B+
80% to 84.99%
B
75% to 79.99%
C+
70% to 74.99%
C
65% to 69.99%
D+
60% to 64.99%
D
<60%
F
Percentage of credit hours completed
(CRPER)
0% to 25%
CR1
26% to 50%
CR2
51% to 75%
CR3
76% to 100%
CR4
(Source: Authors design)
Volume 13 - Issue 74
/ February 2024
269
http:// www.amazoniainvestiga.info ISSN 2322- 6307
We transformed the age into three categories.
Students aged 24 to 30 were transformed as A1,
31 to 36 as A2, and 37 to 42 as A3. The six
subject specialties were given to code from S1 to
S6. The cumulative GPA was transformed as per
the grading criteria employed by the UoH.
Accordingly, cumulative GPA >=95% was
transformed as A+, >=90% as A, >=85% as B+,
>=80% as B, >=75% as C+, >=70% as C, >=65%
as D+, >=60% as D, and <60% as F. The
percentage of credit hours completed, ranging
from 0% to 25%, was transformed as CR1, 26%
to 50% as CR2, 51% to 75% as CR3, and 76% to
100% as CR4.
The final dataset consisted of six non-class
attributes and one class attribute. The non-class
attributes were gender (male or female), age (A1
to A3), specialty (S1 to S6), cumulative GPA
(A+ to F), warnings (0 to 2), and percentage of
credit hours completed (CR1 to CR4). The class
attribute was status (active or dropout).
Finally, we converted the dataset into Weka
software's understandable arff format for data
mining.
Stage 4 – Mine Data
In this stage, we employed the classification
approach of data mining to predict key
determinants of public health master students at
the UoH, Saudi Arabia. Prior to experimentation,
we balanced the data based on the class attribute
(active or dropout) utilizing Weka software’s
resample filter. Before balancing, the dataset's
active and dropout cases were 140 and 10,
respectively. Post-balancing, the active and
dropout cases in the dataset were 75 each. We
conducted experimentation by employing two
tree-based and two rule-based algorithms. In
accordance with Refaeilzadeh et al. (2009), we
did experimentation utilizing ten-fold cross-
validation to make sure that the study findings
were valid. Table 2 portrays the results of the
experiments.
Table 2.
Experimentation Results
Algorithm
Accuracy-%
TPR
FPR
Precision
Recall
F-Measure
J-48
90.67%
0.907
0.093
0.909
0.907
0.907
REP-Tree
86.67%
0.867
0.133
0.869
0.867
0.866
J-Rip
86.00%
0.860
0.140
0.862
0.860
0.860
PART
95.33%
0.953
0.047
0.957
0.953
0.953
Note: The meaning of TPR is true positive rate, whereas FPR is false positive rate.
(Source: Authors design)
The Table 2 experimentation results depict that
the PART algorithm performs better than the
other three algorithms in terms of accuracy-%
(95.33%), TP-rate (0.953), precision (0.957),
recall (0.953), and F-measure (0.953). The FP-
rate (0.047) of PART is also lower than the other
three algorithms. In view of these results, we
selected the PART algorithm to extract rules.
Only the rules corresponding to the dropout
class, having coverage above 5%, and accuracy
over 90% were chosen. Table 3 depicts the
chosen rules.
Table 3.
PART Rules
S.No.
Rule(s)
Coverage
Accuracy
1
IF (Warnings = 1) AND (CumGPA = F) THEN (Class = Dropout)
(36.0/1.0)
24%
97.22%
2
IF (CRPER = CR1) AND (CumGPA = D+) THEN (Class = Dropout)
(21.0/1.0)
14%
95.24%
3
IF (CRPER = CR1) AND (Age = A2) THEN (Class = Dropout)
(18.0/1.0)
12%
94.44%
4
IF (CumGPA = D+) AND (CRPER = CR1) THEN (Class = Dropout)
(15.0/1.0)
10%
93.33%
5
IF (CumGPA = D+) AND (Age = A2) AND (Specialty = S1) THEN
(Class = Dropout) (12.0/1.0)
8%
91.67%
(Source: Authors design)
270
www.amazoniainvestiga.info ISSN 2322- 6307
Rule 1 states that if a student has received one
academic warning and the cumulative GPA
belongs to the F category (means less than 60%
or 2.4), then the student is likely to dropout. This
rule covered 36 out of 150 instances and gave
incorrect result once. Accordingly, the rule
coverage and accuracy are 24% and 97.22%,
respectively.
Rule 2 informs that if the percentage of credit
hours completed by the student belongs to the
CR1 category, (means between 0% to 25%) and
cumulative GPA belongs to D+ category (means
between 65% (2.4) to 69.99% (2.6)), then the
student is likely to dropout. This rule covered 21
out of 150 instances and gave incorrect result
once. Accordingly, the rule coverage and
accuracy are 14% and 95.24%, respectively.
Rule 3 asserts that if the percentage of credit
hours completed by the student belongs to the
CR1 category (means between 0% to 25%) and
the age belongs to the A2 category (means
between 31 and 36 years), then the student will
likely drop out. This rule covered 18 out of 150
instances and gave incorrect result once.
Accordingly, the rule coverage and accuracy are
12% and 94.44%, respectively.
Rule 4 states that if the cumulative GPA belongs
to the D+ category (means between 65% (2.4) to
69.99% (2.6)) and the percentage of credit hours
completed by the student belongs to the CR1
category (means between 0% to 25%), then the
student is likely to dropout. This rule covered 15
out of 150 instances and gave incorrect result
once. Accordingly, the rule coverage and
accuracy are 10% and 93.33%, respectively.
Rule 5 informs that if the cumulative GPA
belongs to the D+ category (means between 65%
(2.4) to 69.99% (2.6)) and age belongs to the A2
category (means between 31 and 36 years), then
the student is likely to dropout. This rule covered
12 out of 150 instances and gave incorrect result
once. Accordingly, the rule coverage and
accuracy are 8% and 91.67%, respectively.
Stage 5 – Evaluate Knowledge
In this stage, we discussed the chosen rules with
the domain experts. The domain experts agreed
with all the five rules. Consequently, we accepted
all the five rules.
Stage 6 – Utilize Knowledge
In this stage, we utilized the five rules to identify
the determinants of health science masters’
students’ dropout at the UoH.
Table 4.
Critical Dropout Determinants
Rule
Age
Specialty
CumGPA
Warnings
CRPER
Class
R-1
F
1
Dropout
R-2
D+
CR1
Dropout
R-3
A2
CR1
Dropout
R-4
D+
CR1
Dropout
R-5
A2
S1
D+
Dropout
Individual
Institutional
Academic
Dropout
(Source: Authors design)
Table 4 depicts that individual, institutional, and
academic determinants affect public health
university student dropout in Saudi Arabia.
Students who have a low cumulative GPA (less
than 2.6) are at risk of dropping out. Students
with a GPA of less than 2.4 and who have
received an academic warning are especially
likely to drop out. The dropout risk is higher in
the case of students who have completed less
than 25% of their credit hours. The freshmen who
have a low GPA and belong to the age group 31
to 36 years are likely to dropout. Also, the low-
performing students aged 31 to 36 years and have
taken the subject specialty of electronic health
could drop out.
Results and discussion
According to the current study findings, student
attrition has multiple causes, such as individual,
institutional, and academic determinants. The
study results show that a student’s age is an
individual determinant of dropout. In this regard,
the current study's findings are consistent with
those of Rodríguez-Muñiz et al. (2019), who
found that a student's age is one of the factors
influencing their dropout. However, the current
study results contribute to the literature by
indicating that low-performing public health
master's students between the ages of 31 and 36
are more likely to drop out.
Volume 13 - Issue 74
/ February 2024
271
http:// www.amazoniainvestiga.info ISSN 2322- 6307
The current study results reveal that the subject
specialty taken by the students could be a
determinant of their dropout. This could be a
significant determinant of dropout for students
aged 31 to 36 whose academic performance is
below average. In this regard, the current study's
findings resonate with Bharadwaj & Pal (2012),
who stated that the subject specialty taken by
students could influence their dropout. However,
the present study contributes to the body of
knowledge by revealing that students between
the ages of 31 and 36 who have taken a subject
specialty in electronic health could drop out.
The present study further informs that student’s
cumulative GPA, academic warnings, and
percentage of credit hours completed are the
academic determinants of their dropout. This
study indicates that students with low academic
performance are at risk of dropping out. This
result is in alignment with Casanova et al. (2018),
Ahmed et al. (2018), Mubarak et al. (2020),
Singh & Alhulail (2022), and Hashim et al.
(2024) who anticipated that students with low
academic performance could dropout. The
current study enriches the body of knowledge by
suggesting that public health master students
with a cumulative GPA of less than 2.60 are at
risk of dropping out. The students could be given
academic warnings due to their low attendance in
class. This result is aligned with Bharadwaj & Pal
(2011) and Rodríguez-Muñiz et al. (2019), who
found that students with low-class attendance
may drop out. The present study also suggests
that low-performing students who have
completed fewer than 25 percent of credit hours
(freshmen) are more likely to dropout. This result
is aligned with Zhang et al. (2010), Casanova et
al. (2018), and Rodríguez-Muñiz et al. (2019),
who found that low-performing freshmen have a
greater probability of dropping out.
The study findings also depict a relationship
between individual, institutional, and academic
determinants of dropout. Age is an individual
determinant of dropout, the subject specialty is
an institutional determinant of dropout, whereas
cumulative GPA, academic warnings, and
percentage of credit hours completed are
academic determinants of dropout. The results of
the study indicate that masters’ of public health
students between the ages of 31 and 36, who are
specializing in electronic health and exhibit low
academic performance in terms of CGPA and
attendance, are more likely to drop out.
Conclusions
The issue of student dropout is a complex issue
due to the long-term effects it can have on a
student's life and career. Student dropout is a
serious issue for Saudi Arabia as it has invested
significant resources to promote its education
sector. Particularly, the dropout of public health
masters’ students is more serious as it reduces the
supply of adequate health educators for the
country. In addition, Saudi Arabia has invested
large resources to upgrade its education and
health sectors as part of its Vision 2030 program
(Saudi Gazette, 2020); hence, Saudi educational
institutions should have dropout prevention
measures in place.
In this study, we employed the data mining’s
classification approach to identify the key
determinants of student dropout in the public
health master's program at the UoH. We adapted
the Cios et al. (2000) model to conduct the data
mining analysis. We analyzed 150 student
records to identify the key determinants of
students’ dropout. The analysis was conducted
using two tree-based (J-48 and Rep-Tree) and
two rule-based algorithms (J-Rip and PART).
The PART method was used to extract the rules
because it performed the best among the four
algorithms. The extracted knowledge informed
that individual (age), institutional (specialty), and
academic (cumulative GPA, academic warnings,
and percentage of credit hours completed)
determinants affect public health university
student dropout in Saudi Arabia.
The study reveals that low-performing students
who have been issued an academic warning are
likelier to drop out. Freshmen with low academic
performance are especially vulnerable to
dropping out of university. The study also
suggests that students between the ages of 31 and
36 who have studied technical courses as a
subject specialization (e.g., electronic health)
may drop out.
To develop dropout prevention strategies,
universities should consider these individual,
institutional, and academic determinants. The
universities should pay close attention to the
academic performance of the freshmen. The
freshmen whose performance is below average
should be offered remedial classes. Further, they
should be given academic counseling to
discourage them from dropping out. The
universities should also monitor the performance
of the students in technical courses (e.g.,
electronic health). The students who perform
below average in the technical courses should
272
www.amazoniainvestiga.info ISSN 2322- 6307
also be given remedial classes to improve their
performance. The universities should increase
the practical content of the technical courses to
improve students' understanding of these
courses.
Limitations and future research
The present study has shortcomings that can be
resolved in future research. The first limitation
pertains to the sample size of 161 public health
master's students. We obtained data from the
UoH, where a sample of 161 students pursuing a
master's degree in public health was available.
Future research can opt for a larger sample size
to enhance the generalizability of the research
findings. The current study exclusively utilized
data from a single public university in Saudi
Arabia. Future research can gather data from
multiple public and private universities in Saudi
Arabia to extrapolate the research results. Future
research can also gather data from universities in
other Gulf Cooperation Council countries due to
their analogous political, economic, and social
characteristics to Saudi Arabia.
The second limitation is the selection of study
variables. The study focused on identifying
factors contributing to student dropout, such as
individual, institutional, and academic variables.
The current research did not consider economic
factors because student economic data from the
UoH was unavailable. Subsequent studies can
gather this data using questionnaires to enhance
the existing research results.
The current study used data mining techniques to
investigate the UoH public health masters’
student dropout rates. The data mining technique
has its advantages but relies on classified data
rather than continuous data. The regression
analysis utilizes continuous data to uncover
subtle insights. Future research can utilize
regression analysis as a robustness check on the
current research findings.
Lastly, the current study focused on master's
students in public health. Future research can
select both undergraduate and graduate students
majoring in public health to extend the research
findings to the field of public health. Future
research can select students from various
disciplines, such as Applied Sciences and
Engineering, to compare dropout profiles and
draw generalized research findings.
Bibliographic references
Abdulghani, H. M., Alanazi, K., Alotaibi, R.,
Alsubeeh, N. A., Ahmad, T., & Haque, S.
(2023). Prevalence of potential dropout
thoughts and their influential factors among
Saudi medical students. SAGE Open, 13(1),
215824402211469.
https://doi.org/10.1177/21582440221146966
Ahmed, U., Umrani, W. A., Qureshi, M. A., &
Samad, A. (2018). Examining the links
between teachers support, academic efficacy,
academic resilience, and student engagement
in Bahrain. International Journal of
Advanced and Applied Sciences, 5(9), 39-46.
https://doi.org/10.21833/ijaas.2018.09.008
Alam, F., Singh, H. P., & Singh, A. (2022).
Economic Growth in Saudi Arabia through
Sectoral Reallocation of Government
Expenditures. SAGE Open, 12(4), 1-13.
https://doi.org/10.1177/21582440221127158
Alazemi, N. F. S. A. (2023). The use of
information technology applications by
faculty members at the college of basic
education in the public authority for applied
education and training in the state of Kuwait.
International Journal of Advanced and
Applied Sciences, 10(3), 136-142.
https://doi.org/10.21833/ijaas.2023.03.018
Alhamad, I. A., & Singh, H. P. (2021). Decoding
Significant and Trivial Factors Influencing
Online Hotel Ratings: The Case of Saudi
Arabia's Makkah City. International
Transaction Journal of Engineering,
Management, & Applied Sciences &
Technologies, 12(7), 12A7H, 1-11.
https://doi.org/10.14456/ITJEMAST.2021.1
34
Alhamuddin, A., Inten, D. N., Mulyani, D.,
Suganda, A. D., Juhji, J., Prachagool, V., &
Nuangchalerm, P. (2023). Multiple
intelligence-based differential learning on
critical thinking skills of higher education
students. International Journal of Advanced
and Applied Sciences, 10(8), 132-139.
https://doi.org/10.21833/ijaas.2023.08.015
Alhulail, H. N., & Singh, H. P. (2023). Impact of
multimedia technology on university students
learning agility and creativity. Amazonia
Investiga, 12(70), 189-199.
https://doi.org/10.34069/AI/2023.70.10.17
Alkhalil, A. (2021). Decision support model to
adopt big data analytics in higher education
systems. International Journal of Advanced
and Applied Sciences, 8(6), 67-78.
https://doi.org/10.21833/ijaas.2021.06.008
Al-Omar, B. (2023). About King Saud
University. King Saud University Accessed
Volume 13 - Issue 74
/ February 2024
273
http:// www.amazoniainvestiga.info ISSN 2322- 6307
December 12, 2023. from
https://tinyurl.com/y5cek94f
Baalmann, T. (2023). Health-Related Quality of
Life, Success Probability and Students’
Dropout Intentions: Evidence from a German
Longitudinal Study. Research in Higher
Education. 65.
https://doi.org/10.1007/s11162-023-09738-7
Bharadwaj, B. K., & Pal, S. (2011). Mining
educational data to analyze students
performance. International Journal of
Advanced Computer Science and
Applications, 2(6).
https://doi.org/10.14569/ijacsa.2011.020609
Bharadwaj, B. K., & Pal, S. (2012). Data Mining:
A prediction for performance improvement
using classification. International Journal of
Computer Science and Information Security,
9(4), 1-5. https://tinyurl.com/mvzvun6v
Casanova, J. R., Cervero, A., Núñez, J. C.,
Almeida, L. S., & Bernardo, A. (2018).
Factors that determine the persistence and
dropout of university students. Psicothema,
30(4), 408-414.
https://doi.org/10.7334/psicothema2018.155
CGIJ. (2023). List of universities and colleges in
Saudi Arabia. Consulate General of India,
Jeddah. Retrieved December 16, 2023, from
http://www.cgijeddah.com/listofuniversity.p
df
Cios, K. J., Pedrycz, W., Swiniarski, R. W., &
Kurgan, L. A. (2007). Data Mining: A
Knowledge Discovery Approach (1st ed.).
New York, NY: Springer US.
https://doi.org/10.1007/978-0-387-36795-8
Cios, K. J., Teresinska, A., Konieczna, S.,
Potocka, J., & Sharma, S. (2000). A
knowledge discovery approach to diagnosing
myocardial perfusion. IEEE Engineering in
Medicine and Biology Magazine, 19(4),
17-25. https://doi.org/10.1109/51.853478
Dlungwane, T., & Voce, A. (2020). Exploring
student persistence to completion in a Master
of Public Health programme in South Africa.
African Journal of Health Professions
Education, 12(1), 17.
https://doi.org/10.7196/ajhpe.2020.v12i1.11
83
Gentle, J. E., Härdle, W. K., & Mori, Y. (Eds.).
(2012). Handbook of Computational
Statistics: Concepts and Methods (2nd ed.,
Springer Handbooks of Computational
Statistics). Heidelberg, Germany: Springer-
Verlag. https://doi.org/10.1007/978-3-642-
21551-3
Han, J., Kamber, M., & Pei, J. (2012). Data
Mining: Concepts and Techniques (3rd ed.).
Waltham, MA: Morgan Kaufmann.
https://doi.org/10.1016/C2009-0-61819-5
Hashim, R. A., Lim, H. E., Jafar, M. F.,
Shanmugam, S. K. S., & Bukhari, N. (2024).
Statistical identification of predictors of
dropout in secondary education: evidence
from Malaysia. Journal of the Asia Pacific
Economy, 1-27.
https://doi.org/10.1080/13547860.2024.2306
673
Ibeaheem, H. A., Elawady, S., & Ragmoun, W.
(2018). Saudi Universities and higher
education skills on Saudi Arabia.
International Journal of Higher Education
Management, 04(02), 1-14.
https://doi.org/10.24052/ijhem/v04n02/art05
Kumari, R., & Singh, H. P. (2022). Role of
Incident Reporting System in Healthcare
Management: A Case of Multispeciality
Tertiary Hospital in India. International
Journal of Information Movement, 6(IX),
12-18. https://tinyurl.com/38pejmwb
Liu, H., Hussain, F., Tan, C. L., & Dash, M.
(2002). Discretization: An Enabling
Technique. Data Mining and Knowledge
Discovery, 6, 393-423.
https://doi.org/10.1023/a:1016304305535
Márquez-Vera, C., Cano, A., Romero, C.,
Noaman, A. Y., Fardoun, H. M., &
Ventura, S. (2015). Early Dropout Prediction
using Data Mining: A Case Study with High
School Students. Expert Systems, 33(1),
107-124. https://doi.org/10.1111/exsy.12135
Mubarak, A. A., Cao, H., & Zhang, W. (2020).
Prediction of students’ early dropout based on
their interaction logs in online learning
environment. Interactive Learning
Environments, 30(8), 1414-1433.
https://doi.org/10.1080/10494820.2020.1727
529
Pal, S. (2012). Mining Educational Data to
Reduce Dropout Rates of Engineering
Students. International Journal of
Information Engineering and Electronic
Business, 4(2), 1-7.
https://doi.org/10.5815/ijieeb.2012.02.01
Refaeilzadeh, P., Tang, L., & Liu, H. (2009).
Cross-Validation. In L. Liu & M. T. Özsu
(Eds.), Encyclopedia of Database Systems
(pp. 532–538). Springer, Boston, MA.
https://doi.org/10.1007/978-0-387-39940-
9_565
Robert, K. J. B. (2023). Faculty commitment and
performance in Montfortian educational
institutions: Basis for a faculty development
program. International Journal of Advanced
and Applied Sciences, 10(2), 113-127.
https://doi.org/10.21833/ijaas.2023.02.015
Rodríguez-Muñiz, L. J., Bernardo, A. B.,
Esteban, M., & Díaz, I. (2019). Dropout and
transfer paths: What are the risky profiles
274
www.amazoniainvestiga.info ISSN 2322- 6307
when analyzing university persistence with
machine learning techniques? Plos One,
14(6), 1-20.
https://doi.org/10.1371/journal.pone.021879
6
Rotem, N., Yair, G., & Shustak, E. (2020).
Dropping out of master’s degrees: objective
predictors and subjective reasons. Higher
Education Research and Development, 40(5),
1070-1084.
https://doi.org/10.1080/07294360.2020.1799
951
Saudi Gazette. (2020, May 20). Full text of Saudi
Arabia’s Vision 2030 | Al Arabiya English.
https://tinyurl.com/4nt5pxan
Singh, A., Singh, H. P., Alam, F., & Agrawal, V.
(2022b). Role of Education, Training, and E-
Learning in Sustainable Employment
Generation and Social Empowerment in
Saudi Arabia. Sustainability, 14(14), 8822.
https://doi.org/10.3390/su14148822
Singh, H. P., & Alhamad, I. A. (2022a).
Influence of National Culture on Perspectives
and Factors Affecting Student Dropout: A
Comparative Study of Australia, Saudi
Arabia, and Ethiopia. Archives of Business
Research, 10(11), 287-300.
https://doi.org/10.14738/abr.1011.13508
Singh, H. P., & Alhamad, I. A. (2022b). A Data
Mining Approach to Predict Key Factors
Impacting University Students Dropout in a
Least Developed Economy. Archives of
Business Research, 10(12), 48-59.
https://doi.org/10.14738/abr.1012.13556
Singh, H. P., & Alhulail, H. N. (2022). Predicting
Student-Teachers Dropout Risk and Early
Identification: A Four-Step Logistic
Regression Approach. IEEE Access, 10,
6470-6482.
https://doi.org/10.1109/access.2022.3141992
Singh, H. P., & Alodaynan, A. M. M. (2023). The
role of educational technology in developing
the cognitive and communicative skills of
university students: A Saudi Arabian case.
International Journal of Advanced and
Applied Sciences, 10(7), 157-164.
https://doi.org/10.21833/ijaas.2023.07.017
Singh, H. P., & Chand, P. (2012). ICT Education:
Challenges and Opportunities. In D. Parimala
(Ed.), Role of Teachers in Changing Context:
Policy and Practice (1st ed., pp. 255–263).
Kanishka Publishers, Distributors.
https://tinyurl.com/3cvexykt
Singh, H. P., Agarwal, A., & Das, J. K. (2013).
Implementation of E-Learning in Adult
Education: A Roadmap. Mumukshu Journal
of Humanities, 5(1), 229-232.
https://tinyurl.com/yfcws7rw
Singh, H. P., Alshallaqi, M., & Altamimi, M.
(2023). Predicting Critical Factors Impacting
Hotel Online Ratings: A Comparison of
Religious and Commercial Destinations in
Saudi Arabia. Sustainability, 15(15), 11998.
https://doi.org/10.3390/su151511998
Singh, H. P., Jindal, S., & Kaurav, R. P. S.
(2011a). Adult Education and E-Learning. In
Proceedings of the National Conference on
Turbulent Business Environment: The Road
Ahead. Rohini, Delhi, India; Gitarattan
International Business School (giBS).
https://tinyurl.com/ru9dhne7
Singh, H. P., Jindal, S., & Samim, S. A. (2011b).
A Critical Study on Adoption of E-Learning
for Development of Human Resources in
Developing Countries. Mumukshu Journal of
Humanities, 3(3), 116-120.
Singh, H. P., Jindal, S., & Samim, S. A. (2011c).
Role of Human Resource Information System
in Banking Industry of Developing
Countries. Special Issue of the International
Journal of the Computer, the Internet and
Management, 19(SP1), 59.1-59.5.
https://bit.ly/3coQmWw
Singh, H., & Alhulail, H. N. (2023). Information
Technology Governance and Corporate
Boards’ Relationship with Companies’
Performance and Earnings Management: A
Longitudinal Approach. Sustainability,
15(8), 6492.
https://doi.org/10.3390/su15086492
Singh, H., Singh, A., Alam, F., & Agrawal, V.
(2022a). Impact of Sustainable Development
Goals on Economic Growth in Saudi Arabia:
Role of Education and Training.
Sustainability, 14(21), 14119.
https://doi.org/10.3390/su142114119
Singh, H.P., & Alhamad, I. A. (2021).
Deciphering Key Factors Impacting Online
Hotel Ratings Through the Lens of Two-
Factor Theory: A Case of Hotels in Makkah
City of Saudi Arabia. International
Transaction Journal of Engineering,
Management, & Applied Sciences &
Technologies, 12(8), 12A8M, 1-12.
https://doi.org/10.14456/ITJEMAST.2021.1
60
Singh, H.P., & Alwaqaa, M. A. M. (2023). The
educational technology's impact on youth
creativity and innovation: A case of Ha’il
region of Saudi Arabia. Amazonia Investiga,
12(66), 144-154.
https://doi.org/10.34069/AI/2023.66.06.14
Yadav, S. K., Bharadwaj, B., & Pal, S. (2012).
Mining Education Data to Predict Student's
Retention: A Comparative Study.
International Journal of Computer Science
Volume 13 - Issue 74
/ February 2024
275
http:// www.amazoniainvestiga.info ISSN 2322- 6307
and Information Security, 10(2), 113-117.
https://tinyurl.com/4jndx6dt
Ye, X., Zhai, M., Feng, L., Xie, A., Wang, W., &
Wu, H. (2022). Still want to be a doctor?
Medical student dropout in the era of
COVID-19. Journal of Economic Behavior
and Organization, 195, 122-139.
https://doi.org/10.1016/j.jebo.2021.12.034
Zhang, Y., Oussena, S., Clark, T., & Kim, H.
(2010). Use data mining to improve student
retention in higher education - A case study.
In ICEIS - 12th International Conference on
Enterprise Information Systems. Retrieved
December 27, 2023, from
http://shura.shu.ac.uk/11970/
