Syntax Literate: Jurnal Ilmiah Indonesia
�p�ISSN: 2541-0849 e-ISSN: 2548-1398
Vol. 7, No. 10, Oktober 2022
PREDICTING
STUDENT PERFORMANCE USING MACHINE LEARNING FOR STUDENT MANAGEMENT IN UNIVERSITY
Berlit
Deddy Setiawan, Dermawan Wibisono
Institut Teknologi Bandung, Indonesia
E-mail: [email protected],
[email protected]
Abstract
Higher
education institutions play an important role in providing quality education
and producing skilled human resources. In Indonesia, the demand for higher
education is increasing due to population growth and increasing awareness of
the importance of higher education. ABC University, which is currently ranked
46-50 in the Indonesia Uni Rank 2023, faces challenges
in ranking. To thrive in this competitive landscape, universities must be
selective in admitting quality students and ensuring an effective academic
development process. Machine learning capabilities can be leveraged to predict
potential student academic performance, facilitate timely interventions, and
support for improving learning outcomes. However, currently there is no
research that focuses on building predictive models that integrate student
profiles with academic achievement. This study aims to establish a relationship
between the theory of the Random Forest algorithm and the prediction of
potential student achievement. The aim is to develop an accurate and efficient
method for managing student affairs at ABC University. This study uses both quantitative
and qualitative approaches, with a focus on numerical data analysis and
produces classification predictions. The research process begins with a
thorough analysis of the business situation to understand the university
environment and determine research topics. The researcher then defines research
boundaries, prioritizes key issues, and develops a research framework. The
study analyzed the profiles of students who graduated in 2016-2017, combined
with academic achievement data. Regression test was conducted to determine the
effect of 18 attributes on performance. Random Forest Machine Learning was
compared with other techniques to identify the most accurate predictive model
for student academic performance. ABC University, the Random Forest model achieves
a prediction rate of 89.60%.
Keywords: Higher education institutions, Predicting students'
potential academic performance, Random Forest algorithm, Factors influencing
student performance
Introduction
Higher
education institutions are academic educational institutions that play a role
in providing relevant and high-quality higher education, capable of producing
qualified human resources that meet the qualifications demanded by the job
market (Tien et al., 2020). That higher education
refers to the university level, consisting of various faculties that offer
academic education in specific disciplines (Barthos, 1992). In Indonesia, there is a
growing demand for higher education due to population growth and increased
awareness of the importance of education (Ajisuksmo, 2017). However, it is worth
noting that as of 2022. Indonesia has 4,593 universities, but only 20 campuses
are included in the world rankings (Lukman & Said, 2022). Of the 20 campuses, only
five are included in the top 500. �Below
is a table of the top five universities in Indonesia, including the ranking of
ABC University (Rosser, 2019).
Table
1
Top
5 University in Indonesia, UniRank version 2023
|
Rank |
University |
City |
|
1 |
Universitas Indonesia |
Depok |
|
2 |
Universitas
Gajah Mada |
Sleman |
|
3 |
Institut Teknologi Bandung |
Bandung |
|
4 |
Universitas Brawijaya |
Malang |
|
5 |
Universitas Bina Nusantara |
Jakarta |
|
46-50 |
ABC University |
- |
Based
on the table 1 above, 4 out of 5 top universities are public universities and
one is a private university. ABC University is ranked 46-50, which is still
quite far behind. There are 10,157,323 senior high school and vocational high
school students in Indonesia. both public and private Those who are active in
the 12th grade or 3rd grade in 2023 have the potential to become university
students (Ramadhan & Megawati, 2023). Below is a table of the
top five regions in Indonesia with the highest number of high school and
vocational school students.
�
Top 5 regions with the
highest number of high school and vocational school students in Indonesia
|
No |
Regions |
Senior High School |
Vocational School |
|
1 |
West
Java Province |
792.478 |
1.066.366 |
|
2 |
East
Java Province |
535.577 |
761.539 |
|
3 |
Central
Java Province |
437.985 |
783.805 |
|
4 |
North
Sumatra Province |
381.827 |
304.346 |
|
5 |
South
Sulawesi Province |
228.913 |
123.12 |
Based
on tables 1 and 2, the competition among prospective students is spread across
the island of Java, with West Java province being the largest market, which
also includes the market for ABC University. The competition among universities
to recruit high-quality students has become increasingly intense (Musselin, 2018). Therefore, each
university is expected to strive to provide the best quality education to
succeed in this competitive environment (Hemsley-Brown et al., 2016). The success of a higher
education institution is often measured by the quality of students who pursue
education at that institution (Dumford & Miller, 2018). Conversely, the failure
to produce high-quality students is often seen as a lack of management
capability on the part of the university in delivering the teaching and
learning process (Markova et al., 2017). Below is a map showing
the distribution of universities in Indonesia (Negoro et al., 2021).
Figure 1
Distribution of High Education
Institutions in Indonesia 2020
Based
on Figure 1, the highest distribution of universities in Indonesia is on the
island of Java to obtain high-quality student outcomes, universities ABC need
to be selective in accepting qualified prospective students and are also
expected to ensure the academic development process of students, providing them
with a high chance of success in their education. One crucial step in the
selection and management of the education process is predicting the potential
academic performance of students (Alyahyan & D�şteg�r, 2020).
Predicting
the potential academic performance of students is essential because by
analyzing student data and indicators, it is possible to assess whether a
student is likely to succeed and perform well in their education or not (Casillas et al., 2012). This prediction will
utilize machine learning capabilities to determine predictions and identify the
most significant attribute indicators that influence student performance. this
enables timely interventions and the necessary support to improve
learning.�
This
prediction serves as one of the considerations for assessing the potential
academic performance and serves as a warning for students with declining or
poor academic performance potential. This prediction will greatly help the
university ABC to remain in the strategy of becoming a university with the best
quality and can compete against other universities, with indicators of the
success of students who excel and manage education with quality. Below is a table
of the study duration classification for undergraduate programs at ABC
University.
Tabel 3
Classification
of Length of Study bachelor�s degree in ABC University
|
No |
Length of Study |
Classification |
|
1 |
3
Years |
Fast |
|
2 |
3.5
� 4 Years |
On
Time |
|
3 |
4.5
� 6 Years |
Lated |
Based
on tables 3 the prediction of students' academic performance will determine
whether students with specific indicators can graduate earlier, graduate on
time, or graduate late. The following is a graph analyzing the "Five
Whys" technique at ABC University.
Figure
2
five
why analysis ABC University
Based
on Figure 2 above, it can be inferred that predicting the potential academic
performance is crucial for ABC University.
Based
on the background that has been described, this study aims to create a
predictive model for potential student academic achievement that can be used in
the Directorate which is responsible for managing students at the university
level.
The
research methodology relates to the steps and procedures that will be carried
out to achieve the objectives and obtain answers to research problems (Hidayat & Alifah, 2022b). These steps and
procedures embody the research framework. This study aims to link the Random
Forest algorithm theory with the method of predicting student performance
potential. The researcher intends to find an accurate and efficient performance
method in improving student management in universities (case study: ABC
University). This study uses two types of models, namely a quantitative
approach and a qualitative approach (Hidayat & Alifah, 2022a). The data originally used
in this study were categorical data and then converted into numerical data. The
converted data was then analyzed using the Decision Tree method to produce
predictions for the potential classification of student academic achievement.
The main research topics and research framework established are supported by a
literature review, which helps identify research gaps to be explored and
provides an understanding of existing research. Researchers also conducted
Focus Group Discussions (FGD) with program managers to consider the factors
that affect student performance and the required predictive output. The next
step involves data collection, where data is collected according to research
needs and stored for processing. The second stage is the data modeling stage
using the implementation of the Decision Tree algorithm and the k-Folds cross
validation technique.
The
population in this study consisted of students at ABC University who would be
the subject of research on predicting student academic achievement.
The sample data to be used includes academic data and student profiles from
2016 to 2017 in all study programs. The primary data for this study were
obtained from the ABC University academic database, supplemented by Focus Group
Discussions (FGD). Interviews were conducted with experts who are competent in
the field to determine the factors that influence student performance which
will be predicted in this study. Data on factors that affect the academic
achievement of ABC University students include taking credits, debt, passing
credits, grade point average (GPA), majors, mother's income, income, father's
income, mother's education, father's education, gender, mathematics, father
Job. This secondary data is the ABC University Academic Database. The method of
data analysis in this study consists of attribute identification.
Identification of research attributes is carried out through a process of
reviewing the literature on previous research as a basis for a preliminary
survey to obtain relevant attributes for modeling and analysis.
Results
and Dicsussion
This
chapter will present the study's results and findings. To examine the data
distribution and analyze the regression lines of independent and dependent
variables, the researcher utilized IBM SPSS Statistics 27 software. The results
and discussion will be divided into several subchapters: Data Distribution,
Implementation of Data Processing using Random Forest and Decision Tree, and
Training and Testing Results.
During the
analysis phase, several steps will be undertaken. These include examining the
data distribution, analyzing
the regression lines of independent and dependent variables, utilizing IBM SPSS
Statistics 27 software, selecting relevant features, and implementing Decision
Tree methodology.
B. Data Distribution
The purpose
of presenting the data distribution in this section is to provide a
comprehensive overview of the distribution of research data. This section
includes the percentage distribution of data for each input variable, as well
as the output variable in the study. The data distribution of the input
attributes from the Science and Technology or Social Humanities Grade 2016 and
2017 datasets is presented. The analysis is performed using IBM SPSS Statistics
27, and the descriptive statistics and frequencies are displayed in the
following table.
Data distribution of input
variable(s)
|
Descriptive Statistics |
|||||||
|
|
N |
Range |
Minimum |
Maximum |
Mean |
Std.
Deviation |
|
|
Gander |
495 |
1 |
1 |
2 |
1.53 |
.500 |
|
|
Admission |
495 |
1 |
1 |
2 |
1.09 |
.293 |
|
|
School Type |
495 |
1 |
1 |
2 |
1.23 |
.420 |
|
|
Mathematic |
495 |
3 |
1 |
4 |
2.86 |
.774 |
|
|
English |
495 |
2 |
2 |
4 |
3.25 |
.564 |
|
|
Father Job |
495 |
5 |
1 |
6 |
2.56 |
1.341 |
|
|
Mother Job |
495 |
5 |
1 |
6 |
4.18 |
1.860 |
|
|
Father Education |
495 |
6 |
1 |
7 |
4.31 |
1.491 |
|
|
Mother Education |
495 |
6 |
1 |
7 |
3.88 |
1.338 |
|
|
Father Income |
495 |
5 |
1 |
6 |
4.22 |
1.099 |
|
|
Mother Income |
495 |
5 |
1 |
6 |
2.25 |
1.620 |
|
|
Debt |
495 |
1 |
1 |
2 |
1.33 |
.469 |
|
|
Length of Study |
495 |
2 |
1 |
3 |
2.06 |
.622 |
|
|
GPA |
495 |
2 |
2 |
4 |
3.68 |
.474 |
|
|
Department |
495 |
1 |
1 |
2 |
1.73 |
.446 |
|
|
Credit Take |
495 |
89 |
114 |
203 |
131.11 |
18.880 |
|
|
Credit pass |
495 |
60 |
114 |
174 |
127.20 |
14.672 |
|
|
%_Attendance |
495 |
38.32 |
37.46 |
75.78 |
48.7414 |
7.83788 |
|
|
Valid N (listwise) |
495 |
|
|
|
|
|
|
From Table 4 above, there are 18 attributes
that will serve as the main data for building the prediction model. These data
will be subjected to linear regression testing in order to determine the most
influential attributes. Here is an example of data distribution:
Table 5
Data distribution of input variable(s)-Gender
|
Gander |
|||||
|
|
Frequency |
Percent |
Valid
Percent |
Cumulative
Percent |
|
|
Valid |
Female |
233 |
47.1 |
47.1 |
47.1 |
|
Male |
262 |
52.9 |
52.9 |
100.0 |
|
|
Total |
495 |
100.0 |
100.0 |
|
|
From Table 5
above, there are 495 student data, consisting of 233 female data, which
accounts for 47.1% of the data distribution, and 262 male data, which accounts
for 52.9% of the total data distribution. The following is a graph representing the data
distribution in Table 5.
Figure
3
Data
distribution Gender
From Figure 3
above, the gender distribution at ABC University exhibits a right-skewed curve
because the number of males is greater than the number of females.
Table 6
Data distribution of input variable(s)-School Type
|
School Type |
|
|||||
|
|
Frequency |
Percent |
Valid
Percent |
Cumulative
Percent |
||
|
Valid |
State |
382 |
77.2 |
77.2 |
77.2 |
|
|
Private |
113 |
22.8 |
22.8 |
100.0 |
||
|
Total |
495 |
100.0 |
100.0 |
|
||
From
Table 6 above, there are 495 student data, consisting of 382 school type is
state, which accounts for 77.2% of the data distribution, and 113 School type
is private, which accounts for 22.8% of the total data distribution. The
following is a graph representing the data distribution in Table IV.3.
Figure 4
Data Distribution School Type.
From Figure IV.2 above, the school type distribution
of students at ABC University exhibits a left-skewed curve because there are
more students from public schools compared to private schools.
C. Business Solution
D. Regression Liner Variable.
Based on the distribution of this data, a regression test will be
conducted to determine which of the 18 attributes influence student performance
and can be utilized as independent variables. The linear regression analysis of
the variables will be performed using IBM SPSS Statistics 27 Software,
employing Automatic Linear Modeling to enhance model
accuracy through boosting. Throughout the analysis, the following attributes will
be generated.
1. Science and
Technology study programs
Figure 5
Model Summary Automatic Linear Regresstion � Science and Technology study programs
From Figure 5 above, the model summary of the
regression results for the Science and Technology study program attribute
indicates an accuracy of 62.6%. The accuracy of the reference model is better
than the accuracy of the ensemble data.
Table 7
Automatic Linear Modeling - enhance model accuracy (boosting) - Science and Technology study
programs
|
No. |
Nodes |
Importance |
Importance |
V4 |
V5 |
|
1 |
Fahter_Edu |
0.0085 |
0.0085 |
Father
Education |
0.0085 |
|
2 |
Math |
0.0139 |
0.0139 |
Mathematic |
0.0139 |
|
3 |
Father_Job |
0.0158 |
0.0158 |
Father
Job |
0.0158 |
|
4 |
School |
0.0248 |
0.0248 |
School
Type |
0.0248 |
|
5 |
Admission |
0.0254 |
0.0254 |
Admission |
0.0254 |
|
6 |
Father_Income |
0.0422 |
0.0422 |
Father
Income |
0.0422 |
|
7 |
GPA |
0.1323 |
0.1323 |
GPA |
0.1323 |
|
8 |
Credit_Pass |
0.1655 |
0.1655 |
Credit
pass |
0.1655 |
|
9 |
Debt |
0.262 |
0.2620 |
Debt |
0.2620 |
|
10 |
Credit_Take |
0.3096 |
0.3096 |
Credit
Take |
0.3096 |
From the table above, it can be observed that out of
the 18 attributes tested, 10 attributes significantly influence the Science and
Technology study programs. The most influential attribute is "Credit_take," with a predictor importance value (V5)
of 0.3096. On the other hand, the attribute with the least influence is
"Father's education," with a predictor importance value (V5) of
0.0085.� Below is a table of the
distribution of attributes with good regression values:
2.
Social Humanities study programs
Figure 6
Model Summary Automatic Linear Regresstion � Social Humanities study programs.
From Figure 6 above, the model summary of the
regression results for the social humanities study program attribute indicates
an accuracy of 55.8%. The accuracy of the reference model is better than the
accuracy of the ensemble data. Below is a table of the distribution of
attributes with good regression values:
Table 8
Automatic
Linear Modeling - enhance model accuracy (boosting) � Social Humanities study
programs
|
No |
Nodes |
Importance |
Importance |
V4 |
V5 |
|
1 |
Father_Job |
0.0076 |
0.0076 |
Father
Job |
0.0076 |
|
2 |
Admission |
0.01 |
0.0100 |
Admission |
0.0100 |
|
3 |
Math |
0.0129 |
0.0129 |
Mathematic |
0.0129 |
|
4 |
Father_Income |
0.0166 |
0.0166 |
Father
Income |
0.0166 |
|
5 |
Gender |
0.021 |
0.0210 |
Gander |
0.0210 |
|
6 |
Mother_Income |
0.0224 |
0.0224 |
Mother
Income |
0.0224 |
|
7 |
GPA |
0.109 |
0.1090 |
GPA |
0.1090 |
|
8 |
Credit_Pass |
0.2217 |
0.2217 |
Credit
pass |
0.2217 |
|
9 |
Debt |
0.2709 |
0.2709 |
Debt |
0.2709 |
|
10 |
Credit_Take |
0.3035 |
0.3035 |
Credit
Take |
0.3035 |
From the table above, it can be observed that out of
the 18 attributes tested, 10 attributes significantly influence the Science and
Technology study programs. The attribute with the highest influence is "Credit_take," which has a predictor importance value
(V5) of 0.3035. On the other hand, the attribute with the least influence is
"Father's job," with a predictor importance value (V5) of 0.0076.
Through the process of Automatic Linear Modeling - enhanced model accuracy (boosting), 12 important
attributes were identified for predicting the academic performance at ABC
University. These attributes include:
Tabel 9
The Attribute Result Regression
|
No. |
Attribute |
Science and
Technology study programs |
Social Humanities
study programs |
|
1 |
Father
Education |
v |
x |
|
2 |
Mathematic |
v |
v |
|
3 |
Father
Job |
v |
v |
|
4 |
School
Type |
v |
x |
|
5 |
Admission |
v |
v |
|
6 |
Father
Income |
v |
v |
|
7 |
GPA |
v |
v |
|
8 |
Credit
pass |
v |
v |
|
9 |
Debt |
v |
v |
|
10 |
Credit
Take |
v |
v |
|
11 |
Gander |
x |
v |
|
12 |
Mother
Income |
x |
v |
From the table above, the
regression analysis resulted in 12 important attributes for Science and
Technology study programs. Gender and mother's income were not included in the
analysis. Similarly, for Social Humanities study programs, father's education
and school type were not
used. These attributes will undergo further analysis using the Random Forest
method and will be compared with other machine learning models. The analysis
will focus on the Random Forest machine learning model.
From the regression analysis conducted on the 12
data attributes, further analysis will be performed using random forest machine
learning. Additional machine learning tests will be conducted to determine
which method yields the highest prediction accuracy. The machine learning
analysis will be executed using Orange Data Mining Software, and the resulting
analysis will be as follows.
3. Science and
Technology study programs
Figure 7
Result Mechine
Learning Science
and Technology study programs
Based on the machine learning
testing conducted on the Science and Technology study programs at ABC
University, it was found that the Random Forest model achieved a prediction
rate of 0.896, an F1 score of 0.889, and a recall of 0.889. The decision tree
structure is as follows:
Figure 8
Decision Tree Science and Technology study
programs
From Figure 8 above, the decision tree results for
Science and Technology study programs yield the following prediction
model: study duration is determined by Credit Semester take. If a student wants
to complete their studies on time, they should take more than 153 Credit
Semester. If they have academic debt, the determination will be based on the
father's occupation.The Random
Forest structure is as follows:
Figure 9
Random Forest Science and Technology study
programs
4.
Social Humanities study programs
Figure 10
Result Mechine
Learning Social Humanities study programs
Based on the machine learning
testing conducted on the Science and Technology study programs at ABC
University, it was found that the Random Forest model achieved a prediction
rate of 0.780, F1 score of 0.780, and recall
of 0.786. The decision tree structure is depicted below:
Figure 11
Decision Tree Social Humanities study programs
From Figure 11 above, the
decision tree results for Science and Technology study programs yield the
following prediction model: study duration is determined by Academic Debt,
Credit Semester Taken, Father's Occupation, Mother's Occupation, and Father's
Income Type. For example, if a student has academic debt, the prediction will
consider the mother's income. If the income type is 1, 2, or 5, the prediction
will be based on the father's occupation. If the father's occupation is 1, 2,
or 4, the prediction will then consider the Credit Semester Taken. If the
credit semester taken is less than 117, the student will be predicted to
graduate late. And the Random Forest structure is as follows:
Figure 11
Random Forest Social Humanities study programs
E. Cross Validation
Based on the machine learning testing results, it is evident that random
forest is the most suitable model for making predictions at ABC University.
Subsequently, a 10-fold Cross Validation test will be conducted
to determine the average prediction results. The Cross Validation analysis will
be performed using Orange Data Mining Software, and the analysis results are
presented below.
1. Science and
Technology study programs
Figure 12
Result 10-folds Cross Validation Science and
Technology study programs
2.
Social Humanities study programs
Figure 13
Result 10-folds Cross Validation Social
Humanities study programs
Based on
the above results, it can be observed that the prediction was conducted using
10-fold cross-validation, yielding the following average results.
Tabel 10
fold cross-validation
|
Model |
F1 |
Prediction |
Recall |
|
Science
and Technology study programs |
0.889 |
0.896 |
0.889 |
|
Social
Humanities study programs |
0.754 |
0.780 |
0.756 |
From the
table above, for the Science and Technology study program model, the prediction
accuracy reaches 89.6%. As for the Social Humanities study program, the
prediction accuracy reaches 78.00%.
F.
Implementation
Plan
1.
Implementation Plan
This academic performance
prediction model can be implemented promptly with the approval of the academic
vice chancellor and the three directorate offices of ABC University. Currently,
ABC University is in the process of developing an early warning system, and the
prediction model will serve as a valuable reference for its functioning.
Figure 14
Early Warning System ABC
University
The proposed framework can be
implemented within the next three months, aligning with the planned timeline
outlined in the following work breakdown structure.
Tabel 11
Implementation Plan
Based
on the issue that occurred at ABC University, from the research results we can
know that: (1) Out of the 18 attributes tested to predict the length of study
for students at ABC University using linear regression, 12 independent
attributes (input variables) were found to be significant. It is worth noting
that these attributes differ from those examined in previous studies. (2) Based
on the analysis conducted using Random Forest on the selected 12 attributes,
the following prediction scores were obtained. (3) Based on the research
findings, it is evident that random forest is the suitable machine learning
model for creating prediction models.
Result Mechine Learning Science and Technology study programs to implement the proposed
model prediction, several considerations should be considered, which include:
(1) Determining the information, data, and reports that need to be supported
and how the academic performance prediction model can complement each other.
(2) Ensuring that the new student management system is effectively communicated
to all directorates responsible for student performance. (3) Analyzing the
benefits and cost implications associated with each activity related to the
implementation of the academic performance prediction model. (4) Providing an
informative and communicative interface accessible to all employees at ABC
University.
BIBLIOGRAPHY
Ajisuksmo, C. R. P. (2017). Practices and challenges
of inclusive education in indonesian higher education. ASEACCU Conference on
�Catholic Educational Institutions and Inclusive Education: Transforming
Spaces, Promoting Practices, and Changing Minds". Assumption University of
Thailand, Bangkok August, 21�27.
Alyahyan, E., & D�şteg�r, D. (2020). Predicting
academic success in higher education: literature review and best practices. International
Journal of Educational Technology in Higher Education, 17, 1�21.
Barthos, B. (1992). Perguruan tinggi swasta di Indonesia:
proses pendirian, penyelenggaraan, dan ujian. (No Title).
Casillas, A., Robbins, S., Allen, J., Kuo, Y.-L., Hanson, M.
A., & Schmeiser, C. (2012). Predicting early academic failure in high
school from prior academic achievement, psychosocial characteristics, and
behavior. Journal of Educational Psychology, 104(2), 407.
Dumford, A. D., & Miller, A. L. (2018). Online learning
in higher education: exploring advantages and disadvantages for engagement. Journal
of Computing in Higher Education, 30, 452�465.
Hemsley-Brown, J., Melewar, T. C., Nguyen, B., & Wilson,
E. J. (2016). Exploring brand identity, meaning, image, and reputation (BIMIR)
in higher education: A special section. In Journal of Business Research
(Vol. 69, Issue 8, pp. 3019�3022). Elsevier.
Hidayat, A. R., & Alifah, N. (2022a). Analysis of The
Basis of The Creative Economy in The Development Strategy of Economic
Innovation. Asian Journal of Social and Humanities, 1(3), 95�104.
Hidayat, A. R., & Alifah, N. (2022b). Reading for
Students in English Language Education Programs. International Journal of
Social Health, 1(2), 57�63.
Lukman, L., & Said, I. M. (2022). Strategi Kesantunan
Pemain Game dalam Saluran Youtube �Jess No Limit.� Jurnal Onoma: Pendidikan,
Bahasa, Dan Sastra, 8(1), 63�76.
Markova, T., Glazkova, I., & Zaborova, E. (2017). Quality
issues of online distance learning. Procedia-Social and Behavioral Sciences,
237, 685�691.
Musselin, C. (2018). New forms of competition in higher
education. Socio-Economic Review, 16(3), 657�683.
Negoro, Y. A. T., Marthanty, D. R., & Soeryantono, H.
(2021). Analysis of the green infrastructure implementation to the enhancement
of environmental support capacity (Case study: Watershed outside University of
Indonesia). IOP Conference Series: Materials Science and Engineering, 1098(2),
22050.
Ramadhan, S., & Megawati, S. (2023). Implementasi
Kebijakan Merdeka Belajar Kampus Merdeka Dalam Meningkatkan Kualitas Pendidikan
Mahasiswa di Universitas Negeri Surabaya. Publika, 1581�1592.
Rosser, A. (2019). Big ambitions, mediocre results: Politics,
power and the quest for world-class universities in Indonesia. Transformations
in Higher Education Governance in Asia: Policy, Politics and Progress, 81�99.
Tien, N. H., Jose, R. J. S., Mai, N. P., Long, N. T., &
Hai, T. V. (2020). Current State of Human Resource in International
Universities of Vietnam. International Journal of Multidisciplinary Research
and Development, 7(7), 22�27.
|
Copyright holder: Berlit Deddy
Setiawan, Dermawan Wibisono (2022) |
|
First publication right: Syntax Literate:
Jurnal Ilmiah Indonesia |
|
This article is licensed under:
|
