Student Performance Prediction via Data Mining & Machine Learning

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 01 | Jan 2022 www.irjet.net p-ISSN: 2395-0072
© 2021, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1080
Student Performance Prediction via Data Mining & Machine Learning
Abhishek Roy1, Akhil Sharma2, Anmol Singh3, Hari Shankar4, Prof. Sahana MP5
1-5Department of Computer Science Engineering, Dayananda Sagar College of Engineering, Bengaluru, Karnataka,
India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Whenever the word ’Education’ is brought into
the limelight, majority of the focus is given towards the
student, it’s a good thing that people are trying to help the
‘Learning’ aspect; but in order to do that, we often forget
about the ‘Teaching’ aspect, helping the teacher(s) in their
method(s) of teaching is equally important as method(s) of
learning.
Key Words: Student, Performance, Prediction, Data
Mining, Machine Learning, Decision Tree, Accuracy
1. INTRODUCTION
Nowadays we face a lot of work when it comes to the
teaching department, a lot of work in the back-end keeps
on going about which the student has no knowledge. And
this is our effort to ease down some work so that the
work-load is a little less in the future. Using a decision tree
algorithm [1][9], the teachers will keep an eye on the
grades of the children and with the help of the student
performance prediction, would be able to keep an eye on
the students who need a little extra attention. One of the
things that we often overlook in the Education System is
the contribution of teachers in helping the next generation
finding the best out of themselves.
2. LITERATURE REVIEW
Huseyin Guruler, Ayhan Istanbullu & Mehmet Karahasan
[1] have said that Data mining is one example of
knowledge discovery, which is used to locate meaningful
and valuable patterns in enormous amounts of data. All of
the tasks involved in the knowledge discovery process are
maintained together with this software system. The
research was based on data collected from university
students.
If the quantity of data and the number of factors are
raised, then reliable forecasts about student ’s progress
can be generated. For various data and challenges, it will
necessitate adjustments and additions.
S. K. Mohamad & Z. Tasir explained [2] the application of
DM in education is currently in its infancy, giving rise to
educational data mining (EDM). EDM emerges as a
paradigm for designing models, activities, procedures, and
algorithms for analyzing educational data. EDM aims to
uncover patterns and create predictions about learners'
behaviors and accomplishments, as well as domain
knowledge material, assessments, educational capabilities,
and applications.The majority of EDM approaches have a
DM profile that is supported by probability, machine
learning, and static disciplines;classification is the most
common task, followed by clustering.In terms of dangers,
they are the impediments to the rational and formal
growth of EDM that prevent, hinder, and obstruct it. As a
result, EDM must address the absence of concrete and
precise theory to build the foundation of how EDM
operates.
C. Anuradha and T. Velmurugan examined the
effectiveness of various decision tree algorithms in terms
of accuracy and processing time[3]. Arpit Trivedi's work
has proposed a simple method for categorizing student
data using a decision tree-based technique. They compiled
a database of 100 students' marks in five subjects for each
of four courses. For implementing measures of a specific
student's type, a frequency measure is employed is used
an extracting features. To create a trained classifier, each
student's most frequent five topic marks are used. They
automatically predicted the class for indefinite students
using a trained classifier.
The First and Second classes were discovered to have had
a major impact on the classifying process. With a bigger
sample dataset, the study might be expanded to look at the
performance of alternative categorization techniques.
Based on certain criteria via student's selection, the rate
of predictions are not consistent amongst algorithms.
V.L. Miguéis et al proposed that Within the study, data
from the first year of a student's educational life (path) is
utilized to propose a two-staged model that employs data
mining methods to predict their eventual academic
accomplishment. [4]. Unlike much educational data mining
literature, educational attainment is typically determined
both from average score received and the duration
required to finish the degree. In addition, this study
suggests dividing it up students depending on the
disparity in either indications of failure or high
performance just at commencement of the degree course
and the model's predicted performance levels. The
research evidence reveals that the proposed model can
properly predict students' quality performance standards
during a preliminary phase in their educational endeavors
with a 95 percent overall accuracy. The approaches may
run into certain technical challenges during the data
extraction and training phases, causing the accuracy rate
to drop by a few percentage points.
Siti Dianah et al made us understand predictive analytics
used in sophisticated analytics,[5] which included machine
learning deployment, to generate high-quality

performance and meaningful data for students at all levels
of education. Most individuals realize that one of the
critical performance factors that teachers should use to
track a student ’s educational development is their grades.
Many different machine learning algorithms have been
proposed in the education arena during the last decade.
Most individuals realize that one of the critical
performance factors that teachers should use to evaluate a
student ’s educational development is their grades.
Pendro Manuel et al states that The emergence of learning
analytics has enabled the construction of predictive
analysis to determine learners' behaviors and actions (e.g.,
performance). Several of these methods, on the other
hand, are just useful in particular learning scenarios, and it
can be tricky to sort to see which factors, which includes
predictor variables and forecast outcome type, impact
predictive findings.[6] Additional details might be helpful
in making generalizations to different scenarios,
comparing strategies, developing prediction models, and
developing more workable alternatives. For concept-
oriented assignments, predictive power was also higher,
and the best models frequently just included the final
encounters. Additionally, it was discovered that multiple-
choice problems were better at predicting than
programming ones. There are certain methodological
flaws to consider. One limitation is the method for filtering
students. Several trainees don't really connect with the
platform and thus should be removed; additional factors
might hold, and the outcomes may well be impacted.
Brijesh Kumar Baradwaj & Saurabh Pal made us realize
that knowledge is buried in the educational data set, but it
can be extracted using data mining techniques. By
providing a data mining model for the institution's higher
education sector, the focus of this research is to show the
strength of data mining techniques inside the context of
teaching and learning. [7] The categorization job works
with a file system to anticipate student segregation based
on historical data. As there are numerous methodologies
for data classification, the tree based method is used here.
Elizabeth Boretz states that the common use of the word
"grade inflation" in the context of higher education is a
possibly harmful exaggeration.[8] Test scores are at an all-
time peak, but research shows that perhaps the surge in
professional learning initiatives and the increased
diversity of academic advising are unrelated. Students are
not customers looking for great marks in exchange for
positive teacher ratings; instead, they want to succeed by
acting together to assist their learning, and colleges and
institutions are up to scratch.
Rathee A, Mathur RP wrote that metadata is a learning of
management and data mining approach for combining
collectively related resources. In the research, there seem
to be a variety of classification algorithms, but decision
tree algorithm are the most widely used since they are
simple to write and grasp especially when compared to
other algorithms. [9] The ID3, C4.5, and CART decision
tree algorithms were utilized to forecast the future of
students' data.
Tarun Verma et all stated that Literacy has long been a
global problem. Every country strives for a 100% literacy
rate. If the literacy rate has improved significantly, there is
still a need to understand the areas where people are still
falling behind. As a result, overall literacy statistics should
be investigated in order to give a quick and correct
framework for assistance inside the management and
planning of educational services,[10] as well as to create
or assist to a data collection, organization, and consuming
platform for education data. Georgia, Cube, Estonia, Latvia,
and Barbados have the highest literacy rates in the
developed world, whereas Mali, South Sudan, Ethiopia,
Niger, and Burkina Faso have the poorest literacy levels.
When it comes to continents, the United States has the
largest literacy level, followed by Europe, Asia, and Africa.
Kerala does have the highest literacy rate (93.9%) in India,
while Bihar has the lowest literacy rate (63.8%). In terms
of area, the South has the greatest literacy rate, followed
by the East, West, and North.
Abdulmohsen Alkushi & Abdulaziz Althewini proposed
that even though student performance predictor can be
used to forecast Saudi student academic [11] achievement,
there are considerable variations between research.
Based on their academic profile, a formula was created to
determine student performance. *continued down*
The survey stands out from others in Saudi Arabia because
it uses a bigger sample size and emphasizes on yearly
student achievement. These variations have added to the
complexity of current research attempts on Saudi
predisposing factors of student performance prediction,
and they should be considered by researchers and policy -
makers alike when making decisions.
Adel M. et al proposed that from 2008-09 to 2010-11
academic years,[12] A retrospective observational
analysis was conducted using information from registers
of enrolled students in the three colleges. The grades
mean, assessing learning rating, and standardized tests
score have been the evaluation's independent variables.
The outcome variable was the mean of the individuals'
first- and second-year grade point averages (GPA).
It wasn't really indicative of all these participants' initial
educational success in KSU's health colleges, according to
the findings of this study.
Hanan Mengash suggested a methodology which was
validated using data from 2,039 students enrolled in a
Saudi public university's Computer Science and
Information College from 2016 to 2019. The findings show
that based on key parameters including high school grade
average,
[13] Prior entrance, individuals' initial university success
can be forecasted using their Academic Aptitude
Admission Test and Associated With particular Test
scores. The data also imply that a child's performance on
the Academic Aptitude Tests is the best predictor of
future achievement.

Mashael A. Al-Barrak and Muna Al-Razgan utilized
everything in their reach to help students improve their
performance and uncover early predictors of their
ultimate grade point average. [14] We used a classification
technique, specifically a decision tree, to predict students'
final GPA based on their grades in earlier classes, which
yielded rather accurate findings.
Mikko Vinni, Wilhelmiina Hämäläinen stated that
intelligent tutoring systems and adaptable learning
environments both require automatic classification.[15]
The learner's current circumstance should be classified
first. This will necessitate the use of a classifier, which is a
model that's used to predict the category worth depends
on other explaining criteria. Teaching method can benefit
from such expectations as well, but computers learning
environments can handle larger class sizes and gather
more information for classifier improvement.
Zlatko J. Kovačić gave an approach in which the [16]
transfer literature has identified scholastic and cultural
variables (age, sex, ethnic origin, marital status) as
potential predictive variables of attrition. Students'
registration forms include the only knowledge, or factors,
we know about them at the time of signing at the Open
Polytechnic of New Zealand.The issue we are attempting
to answer in this work is whether we can predict study
outcomes for freshly enrolled students only based on
enrollment data. The total categorization accuracy of the
CHAID tree was 59.4 percent, while the CART tree was
slightly higher at 60.5 percent, both of which are the
highest accuracy percentages ever obtained.
Ashutosh Nandeshwar, Subodh Chaudhari used student
admissions data to build models that forecast enrollment,
and to evaluate the models using cross-validation, win-
loss tables, and quartile charts. [17] Financial aid was
employed as a controlling factor regardless of their high
school GPA and ACT/SAT results. This component proved
to be the most important in terms of improving the quality
of new students. To see the influence of attributes like
distance from campus and initial point of contact, they
need to be built. The enrollment indicator and the
"persistence indicator" should be combined to uncover
attributes that affect retention.
Diego Garc´ıa-Saiz, Marta Zorrilla proposed that the need
for teachers to forecast their students' performance is
increasing as [18] virtual teaching becomes more popular.
Different machine learning approaches can be utilized to
address this need. In this research, we evaluated the
effectiveness and perception of multiple classification
techniques have been applied to schooling datasets, and
we propose a meta-algorithm to modify the sources of
data and improve the accurateness. The meta-algorithm
given, which uses both Nave Bayes and J48 to analyze and
forecast, produces the best outcomes, with the
precompiled task being completed as of the most
important attribute for the preprocessing technique being
superior.
Pauziah Mohd Arsad et al said that report discusses
advancements in forecasting engineering students'
academic achievement. At the end of semester eight, the
academic [19] achievement was measured using the
cumulative grade point average (CGPA). The research was
carried out at the Universiti Teknologi MARA (UiTM)
Faculty of Electrical Engineering in Malaysia. This model is
limited to electrical degree students at the Faculty, but it
can be extended to other departments with the addition of
appropriate input variables. For each student's software,
the designer must determine the input predictor variables.
Cristóbal Romero et al said that The aim of using internet
discussion discussion boards is only to see how the
choosing of cases and qualities, by use of various
classifiers, and the schedule when data are collected
impact predictive performance and conciseness is to see if
making a prediction accuracy at the completion of the
term and an accurate warning well before end of the
program is relevant. [20] Producing better quality when
just the messages relevant to the topic are used. Without
grouping and connection criteria, nevertheless, we cannot
obtain reliable and generally easy to interpret models.
Farshid Marbouti et al proposed that it is possible to[21]
identify at-risk pupils early and notify both the teachers
and the students using predictive modeling tools. But due
to the unpredictability of students' behavior, it is
impossible to create a 100 percent accurate model of their
performance.
G.Gray, Colm McGuinness, P. Owende’s study looked at
psychometric factors that may be tested early after
enrollment, such as personality, motivation, and learning
techniques. Model accuracy was assessed via cross
validation,[22] and the results were compared to
outcomes when the models were used in a following
academic year. When just under 21 pupils were used to
train the models, they all improved in accuracy.
Ahmed Mueen et al proposed that three classifiers'
prediction performance is assessed and compared. This
research will assist teachers in helping students enhance
their academic performance. The Nave Bayes classifier
exceeds the other two by achieving an overall prediction
accuracy of 86%. [23] The pupils were not interested in
using the forum because there were no assigned marks for
doing so.
Gökhan Akcapinar et al's analyses were carried out using
the Orange data mining tool, and the models were
assessed using ten-fold [24] cross-validation. The
classification model correctly predicted 22 of the 27 failing
students (81.5%) and 45 of the 49 passing students (91.8
percent ).
In another paper, a maximum of 76 2nd undergraduate
enrolled students in a Computer Hardware program
participated in the study. [26] The research attempts to
know two fundamental questions by assessing different
classification models and pre-processing techniques:
which methods and attributes largely determine students'
edge school achievement, and whether academic

achievement can be anticipated in sooner weeks to use
these functionalities and the chosen algorithm.
George Siemens, Ryan S J.d. Baker proposed that Two
research communities have formed to address the
demand: Educational Data Mining (EDM) and [25]
Learning Analytics and Knowledge Management (LAK).
This study supports for improved and formalized
communication and cooperation across both groups in
order to share information, processes, and technologies
for data mining and analysis in the purpose of increasing
both the LAK and EDM domains. Many modern data
mining/analytics solutions, however, need not directly
endorse this goal.
M.I. López et al proposed that in order to see if student
engagement in the course forum is a good predictor of
final course grades, and to see if the proposed clustering
approach can achieve equivalent accuracy to standard
classification algorithms. The EM clustering approach
achieves a level of accuracy that is similar to classic [27]
classification algorithms once more. For educators,
manually evaluating messages is a tough and time-
consuming operation.
Jiechao Cheng stated that Data mining is a strong
analytical tool for enterprises to improve decision-making
and analyze new patterns and relationships, and EDM [28]
includes techniques such as data mining, statistics, and
machine learning. However, the costs and challenges of
deploying EDM applications are high.
Ryan Shaun Baker, Paul Salvador Inventado proposed that
these strategies are used by researchers and practitioners
to investigate new constructs and answer new research
questions. A variety of tech is used in the process of data
collection for data mining [29]. Every bit of data created
requires its own preservation and maintenance. As a
consequence, the installation cost could increase.
Furthermore, an expert must be hired for tooling and
other operations, which will increase the entire price.
Chong Ho Yu et al stated that an internally produced
exam, according to this analysis, can provide [30] arguably
more accurate information about a student's cognitive
abilities. Because the data was taken from a single
institution's data warehouse, the findings cannot be
applied to a broader, national scale until more replication
studies are completed.
Parneet Kaur et al study aids institutions in identifying
pupils who are slow learners, which may then be used to
determine additional [31] assistance for them. EDM is still
in its infancy, but it has a lot of educational potential. The
precision was attained using the Multi Layer Perception
classifier, nevertheless it was only 75%.
Zahyah Alharbi et al proposed that the most important
part is to identify weak pupils who are at risk of receiving
a lower grade or dropping out of school. Designers
recognise circuits that are affiliated with good & evil
results for children with similar traits and academic ability
documents when [32] device choices are accessible,
because when component options are usable, those
components can be proposed or disheartened for students
of similar character traits and academic ability records. In
this case, the solution may be a decision support system
that evaluates the paths and successes of like students to
advise what could have been the top choices for a specific
student. We may also look at how module dependencies
and relationships are measured.
In a paper, E. Smith and P. White suggested to investigate
the qualities of candidates to multiple disciplines, as well
as to function each topic plays in influencing the likelihood
of finishing with a 'outstanding' gpa Despite major and
within variance in graduate results, multiple regression
analysis between educators' social and academic features
and university accomplishment disclosed that the topic
educators researched had some predictive power in
predicting future final degree categories upon attempting
to control for cultural class and previous attainment. This
finding has ramifications for main element at increasing
the number[33] and level of STEM grads in a profession
that is frequently known to as a "shortage" or "priority."
D. KABAKCHIEVA,K. STEFANOVA, V. KISIMOV’s work talks
about specific goal is to uncover interesting patterns in the
[35] available data that can help forecast student
performance at university based on their personal and
pre-university characteristics. The goal of the data mining
project is to forecast student university performance
based on personal and pre-university factors.
Olugbenga Adejo and Thomas Connolly The concept
behind this framework is to use a holistic strategy to
accurately and efficiently anticipate student performance.
The six variable domains that have a significant impact on
student performance will be used in the performance
prediction framework provided. [34]
• Self-efficacy, achievement, goal, and interest are
examples of psychological domains.
• Exam score, presenting ability, and intellectual
competence are all part of the cognitive domain.
• Motivation, learning style, study time, habit, ICT skill, and
online activities are all part of the personality domain.
• Income, income distribution status, parent financial
position, and employment status are all part of the
economic sphere.
• Age, gender, location, ethnicity, marital status, and
disability are all demographic domains. • Course
programme, learning environment, institutional support,
and course workload are all part of the institutional
domain.
Liang Zhao et al conducted an experiment which is
undertaken based on an actual university database of
college kids, which combines multisource behavioral data
from multiple sources, including not just physical and
digital education, but also within even outside the
behaviors in the classroom Performance measures
assessing linear and nonlinear behaviors (e.g., regularity

and stability) of school grounds ways of life are
approximated in [36] to obtain in-depth understanding
through into functionalities that ultimately led to superb
or underperformance; likewise, includes able to represent
variability in chronological lifestyle patterns are derived
utilizing long poor memory (LSTM). (2) Secondly, machine
learning-based classifications approaches are being
proposed to estimate academic achievement. (3) Lastly,
tangible advice is being produced to assist the students
(especially at-risk students) in improving their university
connections and achieve a good study-life mix.
3. CONCLUSION
In this paper, we went through different methodologies,
techniques and ways in which the Student Prediction
Model can be made, trained & tested for the accuracy.
Some statistical and neurological factors were also taken
into consideration and using that it helped in carving a
path of combining few ideas or modifying the existing one
for a better method.
By far, through reading and understanding the case
studies and the research papers, we can possibly conclude
that decision tree method is utilised in this because there
are several ways to data classification. To anticipate the
performance of the student at the end of term, information
such as attendance, class test, seminar, and assignment
marks were gathered from the student's previous
database.
ACKNOWLEDGEMENTS
We would like to express our gratitude towards Dr.
Vindhya P Malagi, Dr. Ramya R S and Prof. Sahana MP for
their guidance and advice(s). We would also like to show
gratitude to our Computer Science Department HOD, Dr.
Ramesh Babu D R and all the teachers/mentors who
helped us understand and get in detail with this project.
REFERENCES
[1] Guruler, H., Istanbullu, A. & Karahasan, M. (2010). A
new student performance analyzing system using
knowledge discovery in higher educational databases.
Computers & Education, 55(1), 247-254. Elsevier Ltd.
Retrieved January 17, 2022 from
https://www.learntechlib.org/p/66621/.
[2] S. K. Mohamad and Z. Tasir, ``Educational data mining:
A review.'' Procedia Social Behav. Sci., vol. 97, pp.
320324, Nov. 2013.
[3] Anuradha, C & Thambusamy, Velmurugan. (2015). A
Comparative Analysis on the Evaluation of
Classification Algorithms in the Prediction of Students
Performance. Indian Journal of Science and
technology. 8. 974-6846.
10.17485/ijst/2015/v8i15/74555.
[4] V. L. Miguéis, A. Freitas, P. J. V. Garcia, and A. Silva,
``Early segmentation of students according to their
academic performance: A predictive modeling
approach,'' Decis. Support Syst., vol. 115, pp. 3651,
Nov. 2018.
[5] Bujang, Siti & Selamat, Ali & Ibrahim, Roliana &
Krejcar, Ondrej & Herrera-Viedma, Enrique & Fujita,
Hamido & Ghani, Nor. (2021). Multiclass Prediction
Model for Student Grade Prediction Using Machine
Learning. IEEE Access. PP. 1-1.
10.1109/ACCESS.2021.3093563.
[6] Moreno-Marcos, Pedro & Pong, Ting-Chuen & Merino,
Pedro & Delgado-Kloos, Carlos. (2020). Analysis of the
Factors Influencing Learners’ Performance Prediction
With Learning Analytics. IEEE Access. PP. 1-1.
10.1109/ACCESS.2019.2963503.
[7] Baradwaj, Brijesh & Pal, Saurabh. (2011). Mining
Educational Data to Analyze Students' Performance.
International Journal of Advanced Computer Science
and Applications. 2. 63-69.
10.14569/IJACSA.2011.020609.
[8] Boretz, Elizabeth. (2004). Grade Inflation And The
Myth Of Student Consumerism. College Teaching. 52.
42-46. 10.3200/CTCH.52.2.42-46.
[9] A. Rathee, R. Mathur (2013) Survey on Decision Tree
Classification algorithms for the Evaluation of Student
Performance(Published in BIOINFORMATICS 8 March
2013)
[10] Tarun Verma, Sweety Raj, Mohammad Asif Khan,
Palak Modi (2012) Literacy Rate Analysis | The
research paper published by IJSER journal is about
Literacy Rate Analysis 1 | ISSN 2229-5518
[11] Alkushi, Abdulmohsen & Althewini, Abdulaziz. (2020).
The Predictive Validity of Admission Criteria for
College Assignment in Saudi Universities: King Saud
bin Abdulaziz University for Health Sciences
Experience. International Education Studies. 13. 141.
10.5539/ies.v13n4p141.
[12] Ability of admissions criteria to predict early academic
performance among students of health science
colleges at King Saud University, Saudi Arabia by Adel
M. Alhadlaq, PhD; Osama F. Alshammari, BDS; Saleh M.
Alsager, PhD; Khalid A. Fouda Neel, MD; Ashry G.
Mohamed, DrPh (2015) Ability of admissions criteria
to predict early academic performance among
students of health science colleges at King Saud
University, Saudi Arabia, J Dent Educ. 2015
Jun;79(6):665-70. PMID: 26034031
[13] Mengash, Hanan. (2020). Using Data Mining
Techniques to Predict Student Performance to
Support Decision Making in University Admission
Systems. IEEE Access. PP. 1-1.
10.1109/ACCESS.2020.2981905.
[14] Al-Barrak, Mashael & Al-Razgan, Muna. (2016).
Predicting Students Final GPA Using Decision Trees: A

Case Study. International Journal of Information and
Education Technology. 6. 528-533.
10.7763/IJIET.2016.V6.745.
[15] Wilhelmiina Hämäläinen & Vinni, Mikko. (2010).
Classifiers for Educational Data Mining.
10.1201/b10274-7.
[16] Kovacic, Zlatko. (2010). Early Prediction of Student
Success: Mining Students Enrolment Data. 647-665.
10.28945/1281.
[17] Nandeshwar, Ashutosh & Chaudhari, Subodh. (2009).
Enrollment Prediction Models Using Data Mining.
[18] García-Saiz, Diego & Zorrilla, Marta. (2011).
Comparing classification methods for predicting
distance students' performance. Journal of Machine
Learning Research - Proceedings Track. 17. 26-32.
[19] Mohd Arsad, Pauziah & Buniyamin, Norlida & Ab
Manan, Jamalul-Lail. (2013). A neural network
students' performance prediction model (NNSPPM).
1-5. 10.1109/ICSIMA.2013.6717966.
[20] Romero, Cristóbal & López, Manuel-Ignacio & Luna,
José María & Ventura, Sebastian. (2013). Predicting
Students’ Final Performance from Participation in On-
line Discussion Forums. Computer Education. 68. 458-
472. 10.1016/j.compedu.2013.06.009.
[21] Marbouti, Farshid & Diefes-Dux, Heidi & Madhavan,
Krishna. (2016). Models for early prediction of at-risk
students in a course using standards-based grading.
Computers & Education. 103.
10.1016/j.compedu.2016.09.005.
[22] Gray, Geraldine & McGuinness, Colm & Owende, Philip.
(2014). An application of classification models to
predict learner progression in tertiary education.
Souvenir of the 2014 IEEE International Advance
Computing Conference, IACC 2014. 549-554.
10.1109/IAdCC.2014.6779384.
[23] Mueen, Ahmed & Zafar, Bassam & Manzoor, Umar.
(2016). Modeling and Predicting Students' Academic
Performance Using Data Mining Techniques.
International Journal of Modern Education and
Computer Science. 11. 36-42.
10.5815/ijmecs.2016.11.05.
[24] Gökhan Akcapinar, Arif Altun, Petek Askar(6 August
2015)Modeling Students’ Academic Performance
Based on Their Interactions in an Online Learning
Environment, SOSED Holistic Education Consultancy
& Publications
[25] Siemens, George & Baker, Ryan. (2012). Learning
analytics and educational data mining: Towards
communication and collaboration. ACM International
Conference Proceeding Series.
10.1145/2330601.2330661.
[26] Akçapınar, G., Altun, A. & Aşkar, P. Using learning
analytics to develop early-warning system for at-risk
students. Int J Educ Technol High Educ 16, 40 (2019).
https://doi.org/10.1186/s41239-019-0172-z
[27] Lopez, M.I. & Luna, José María & Romero, Cristóbal &
Ventura, Sebastian. (2012). Classification via
clustering for predicting final marks based on student
participation in forums. Proc. of 5th Int. Conf. on
Educational Datamining. 148-151.
[28] Cheng, Jiechao. (2017). Data-Mining Research in
Education.
[29] Baker, Ryan & Inventado, Paul. (2014). Educational
Data Mining and Learning Analytics. 10.1007/978-1-
4614-3305-7_4.
[30] Yu, Chong Ho & Yu, Samuel & Digangi, Angel &
Jannasch-Pennell, Charles & Kaprolet,. (2010). A Data
Mining Approach for Identifying Predictors of Student
Retention from Sophomore to Junior Year. Journal of
Data Science. 8. 307-325.
10.6339/JDS.2010.08(2).574.
[31] Kaur, Parneet & Singh, Manpreet & Josan, Gurpreet.
(2015). Classification and Prediction Based Data
Mining Algorithms to Predict Slow Learners in
Education Sector. Procedia Computer Science. 57.
500-508. 10.1016/j.procs.2015.07.372.
[32] Alharbi, Zahyah & Cornford, James & Dolder, Liam &
Iglesia, Beatriz. (2016). Using data mining techniques
to predict students at risk of poor performance. 523-
531. 10.1109/SAI.2016.7556030.
[33] Smith, Emma & White, Patrick. (2014). What makes a
successful undergraduate? The relationship between
student characteristics, degree subject and academic
success at university. British Educational Research
Journal. 41. 10.1002/berj.3158.
[34] Adejo, Olugbenga & Connolly, Thomas. (2017). An
Integrated System Framework for Predicting
Students' Academic Performance in Higher
Educational Institutions. International Journal of
Computer Science and Information Technology. 9.
149-157. 10.5121/ijcsit.2017.93013.
[35] Kabakchieva, Dorina & Stefanova, Kamelia & Kisimov,
Valentin. (2011). Analyzing University Data for
Determining Student Profiles and Predicting
Performance.. 347-348.
[36] L. Zhao et al., "Academic Performance Prediction
Based on Multisource, Multi Feature Behavioral Data,"
in IEEE Access, vol. 9, pp. 5453-5465, 2021, doi:
10.1109/ACCESS.2020.3002791.

Student Performance Prediction via Data Mining & Machine Learning

More Related Content

Similar to Student Performance Prediction via Data Mining & Machine Learning (20)

More from IRJET Journal (20)

Recently uploaded (20)

Student Performance Prediction via Data Mining & Machine Learning