Shalini Kumari et al, International Journal of Computer Science and Mobile Computing, Vol.7 Issue.12, December- 2018, pg.

158-162

Available Online at [Link]

International Journal of Computer Science and Mobile Computing

A Monthly Journal of Computer Science and Information Technology

ISSN 2320–088X
IMPACT FACTOR: 6.017

IJCSMC, Vol. 7, Issue. 12, December 2018, pg.158 – 162

Student Performance Prediction

Techniques: A Review
Shalini Kumari
Research Scholar, Bahra University, Waknaghat, Shimla Hills
raghuwanshishalini186@[Link]

Parul Gazta
Assistant Professor, Bahra University, Waknaghat, Shimla Hills
parulgazta@[Link]

Abstract: The prediction analysis is the approach which can predict future possibilities based on the current information. The
prediction analysis can be done using the technique of classification and neural networks. Every educational institute aims at delivering
quality education to their students, to meet this institute must able to evaluate teachers’ as well as students’ performance so that they
can provide appropriate guideline to student and can able to arrange proper training for teachers also. Many researchers have
developed systems which able to evaluate students’ performance but improving students’ performance is not the sufficient to provide
quality education as teacher plays an important role in educating student.

KEYWORDS: Student performance, prediction, Classification

Introduction

There is no limit on the data storage; the data stored is so large that it is almost impossible to fetch relevant and important
data on time. The data can be stored in files, hard disks, CD drives, databases and several other applications. The fetching of
relevant and important file from this infinite amount of data is called Data Mining. Educational Data Mining is the very new
and advanced method being employed by the data mining and Knowledge Discovery in Databases (KDD). Its main objective
is to focus mainly on the useful patterns and recognize the useful and relevant data from the educational information systems,
like admission system, registration, course management, syllabus management and many other systems. These systems and
projects deals with the students in different stages of educational institutions like schools, colleges and universities [1].
Researchers are working in this field to discover and determine the relevant knowledge to help the educational fields to
manage their smooth working of schools, colleges and universities. This will also enhances the better and improved

© 2018, IJCSMC All Rights Reserved 158

 Shalini Kumari et al, International Journal of Computer Science and Mobile Computing, Vol.7 Issue.12, December- 2018, pg. 158-162

functioning of the institution by managing their activities. It studies the students‟ data and information by classifying the
student‟s data with the help of decision tree or association rules. It is the better and interesting field of research which most
commonly focuses on the observation and understanding of student‟s data which is related to their educational field [2]. This
involves the analyzing the performance data, managing the proper functioning of the educational institutes. It makes sure that
all the students are getting proper and sufficient facilities and they are not facing any problem in their education. It also
includes the classification, in which is the most common data mining technique used to classify the educational institutions
according to their performance. Educational Data Mining (EDM) is used in this research to study the collected student‟s
information by a survey and classification is performed on the basis of the gathered data [3]. This classification, classify and
predicts the student‟s performance in their coming semesters or upcoming examinations. The main objective of this research
is identifying the relationship between the student‟s personal and social factors as well as in their academic performance
analysis. This provides facility to the student and the institutions to work in an organized and synchronized manner, by
analyzing the performance of every student and then they decide who is underperforming and who is performing well in their
academics. This will help the educational institutions to work in a very proper and uniform manner [4]. Every educational
institute aims at delivering quality education to their students, to meet this institute must able to evaluate teachers‟ as well as
students‟ performance so that they can provide appropriate guideline to student and can able to arrange proper training for
teachers also. Many researchers have developed systems which able to evaluate students‟ performance but improving
students‟ performance is not the sufficient to provide quality education as teacher plays an important role in educating
student. So, the system can able to evaluate students‟ as well as teachers‟ performance and it also recommend respective
training to them. Every educational field has the main objective to provide better and advanced knowledge to each and every
student. These institutions are using developed and updated modes of education like whiteboards, projectors, smart study and
many other modes are being employed in the institutions [5]. Various researchers are working on this field and they have
evaluated the student‟s performance but the improved performance of the student is not sufficient to have a proper
functioning of the institutions. They also need to develop such technology which can analyze and manage the complete
working of the educational industries. Therefore, the proposed system helps to study the student‟s as well as teachers overall
performance and suggest proper training of the teachers. This system architecture has sub-components. Database contains the
list of students, teachers and the type of courses the particular institution is offering. It is further divided into three categories
that is; student information, teacher information and the course information. Various data mining techniques have been
employed for the classification like decision tree algorithm [6], support vector machines, artificial neural network and
discriminant analysis. This module makes use of Apache PredictionIO Machine Learning Server, in order to predict and
evaluate the performance. Within any system, the most common components involved as users. The individuals that are
going to utilize the system and for whom they are using it can be defined here. Mainly, the performance of teachers and
students are predicted and evaluated in the designed model. Thus, the education quality, performance as well as the overall
results is improved by taking required actions. The first user of system is known to be the students. The performance of
students is evaluated and predicted by the system. Thus, the individual performance of students can be checked and it can
also be evaluated by them whether they can perform well in exams or not. The system is largely affected by the teachers. The
result that is generated by prediction tool is viewed by the teachers. The performance of students is improved and they are
helped in passing the course by the teacher by making appropriate decisions. The class-wise results of students are viewed by
the administrator or principal. Notices are arranged for teachers as per these results [7]. Also the course that needs to be
offered to students can be decided by this administrator.

Literature Review

Jie Xu, [Link] (2016) proposed a new machine learning technique using which the performance of students in degree programs
is predicted. There are two important features included within the proposed technique [8]. The multiple base predictors are
included to develop a bi-layered structure which is the first feature of proposed method. Also, on the basis of the evolving
performance states of students, a cascade of ensemble predictors is designed using which predictions can be done. For
discovering the course relevance, a data-driven technique is designed as a second feature. It uses probabilistic matrix and
latent factor models. Simulations are performed on a dataset collected at UCLA over three years and the results are achieved
which show that the performance of proposed approach is better as compared to existing techniques.

© 2018, IJCSMC All Rights Reserved 159

 Shalini Kumari et al, International Journal of Computer Science and Mobile Computing, Vol.7 Issue.12, December- 2018, pg. 158-162

S. M. Merchán, [Link] (2016) proposed a predictive model to be applied to predict the academic performance of students.
Several data mining methods are applied on the data of 932 students of a university of Columbia to evaluate and analyze their
performances [9]. On the basis of input data given, the expected results and output characterization and other factors, the
evaluation of results achieved is done. The prediction accuracy is an important parameter to evaluate the performances as
well. Considering the specific details of the population examined and the requirements specified by the institution, the said
pertinence is evaluated. For preventing any kinds of academic risk and desertion, timely decisions are considered important
along with the accompaniment of students with their learning procedure. This research work is developed further by
providing few recommendations and thoughts by different researchers.

Ishwank Singh, [Link] (2016) proposed a simple clustering analysis through which the behavior of student is understood. To
understand if there is a regular improvement in the performance of student, a good benchmark has been set up by the data
mining algorithm [10]. During the admission and placement process, this analysis is very helpful. The projects, internships,
skill sets, Xth, XIIth, and [Link] marks are few parameters included for this analysis. Since the implementation is easy and
the computational efficiency is high, K-means algorithm is used in clustering. In future, other clustering techniques can be
applied to improve efficiency levels. Also, to achieve a better student performance analysis, the ranking or classification of
objects present within the clusters can be done.

Ms. Tismy Devasia, [Link] (2016) proposed classification within the information of student such that on the basis of previously
existing information, the division of students can be predicted [11]. Naïve theorem is applied since several techniques are
used for knowledge classification within the area unit. For the prediction of performance at the top of that particular semester,
various types of information were collected from the previous information of the students available. To encourage the
students of different categories to perform well, the lecturers and students can be benefitted through this study. The students
who need any special guidance can be highlighted through this study. Also, the failure ratio can be reduced with the help of
this. For the upcoming semester examination, acceptable actions can be taken through this.

Yuni Yamasari, [Link] (2016) proposed feature extraction techniques in this paper. The student data was gathered in a serious
game to perform extraction on the basis of category and Bloom‟s Taxonomy [12]. The proposed approach is implemented on
this data to perform certain evaluations. It is seen that the level of accuracy is improved and the execution time is minimized
through this method. In comparison to traditional FCM, the level of accuracy is improved up to 2.3-4.7%. Also, in
comparison to traditional approach, the execution time is 2.2-2.7 seconds faster for the proposed FCM approach. The
performance of clustering process on the achievement of student is enhanced when the features are extracted using
CBE_FCM and BTBE_FCM. Weight is added to each feature for improving the proposed methods. The correlation level of
student achievement is considered to be an important factor here.

Nurul „Ulyani, [Link] (2017) presented that the major factor that leaves a huge impact on the behavioral intentions of student is
the service quality performance. Within seven Malaysian public and private universities the paper-and-pencil questionnaires
were distributed [13]. The descriptive statistics and covariance-based structural equation modeling were used to analyze the
data. The least likely execution of favorable behavioral intentions was influenced by the freedom, serenity, management
dimensions as well as aesthetic factors. A positive behavior towards the student housing was seen as per the results achieved
when students adapted to live in multi-cultural community in which they would have access to good hospitality, personal
privacy and appropriate building ambiance.

© 2018, IJCSMC All Rights Reserved 160

 Shalini Kumari et al, International Journal of Computer Science and Mobile Computing, Vol.7 Issue.12, December- 2018, pg. 158-162

Author’s Names Year Description Outcomes

Jie Xu, Kyeong Ho 2016 A new machine learning technique is Simulations are performed on a dataset
Moon, and Mihaela proposed using which the performance of collected at UCLA over three years and
van der Schaar students in degree programs is predicted. the results are achieved which show that
the performance of proposed approach is
better as compared to existing techniques.
S. M. Merchán, J. 2016 A predictive model to be applied to predict the On the basis of input data given, the
A. Duarte academic performance of students is expected results and output
proposed. Several data mining methods are characterization and other factors, the
applied on the data of 932 students of a evaluation of results achieved is done.
university of Columbia to evaluate and The prediction accuracy is an important
analyze their performances parameter to evaluate the performances as
well.
Ishwank Singh, A 2016 A simple clustering analysis is proposed Since the implementation is easy and the
Sai Sabitha, Abhay through which the behavior of student is computational efficiency is high, K-
Bansal understood. To understand if there is a regular means algorithm is used in clustering.
improvement in the performance of student, a
good benchmark has been set up by the data
mining algorithm.
Ms. Tismy 2016 Classification is proposed within the The students who need any special
Devasia, information of student such that on the basis guidance can be highlighted through this
[Link] T P, of previously existing information, the study. Also, the failure ratio can be
[Link] Hegde division of students can be predicted. reduced with the help of this. For the
upcoming semester examination,
acceptable actions can be taken through
this.
Yuni Yamasari, 2016 Feature extraction techniques are proposed in The performance of clustering process on
Supeno M. S. this paper. The student data was gathered in a the achievement of student is enhanced
Nugroho, I N. serious game to perform extraction on the when the features are extracted using
Sukajaya, Mauridhi basis of category and Bloom‟s Taxonomy CBE_FCM and BTBE_FCM. Weight is
H. Purnomo added to each feature for improving the
proposed methods.
Nurul „Ulyani, 2017 The descriptive statistics and covariance- A positive behavior towards the student
Mohd Najib, based structural equation modeling were used housing was seen as per the results
Nor‟Aini Yusof, to analyze the data. The least likely execution achieved when students adapted to live in
Amin Akhavan of favorable behavioral intentions was multi-cultural community in which they
Tabassi influenced by the freedom, serenity, would have access to good hospitality,
management dimensions as well as aesthetic personal privacy and appropriate building
factors. ambiance.

Conclusion

The prediction analysis is the approach which can predict future possibilities from the current information. The prediction
analysis can be done with the techniques of classification. This review paper is based on the student performance prediction.
The various classification techniques are reviewed in paper for the student performance prediction.

© 2018, IJCSMC All Rights Reserved 161

 Shalini Kumari et al, International Journal of Computer Science and Mobile Computing, Vol.7 Issue.12, December- 2018, pg. 158-162

References
[1] Amin Zollanvari, Refik Caglar Kizilirmak, Yau Hee Kho, and Daniel Hernandez-Torrano, “Predicting Students‟ GPA and
Developing Intervention Strategies Based on Self-Regulatory Learning Behaviors”, 2017, IEEE

[2] Sneha Chandra, Maneet Kaur,” Enhancement of Classification Accuracy of our Adaptive Classifier using Image
Processing Techniques in the Field of Medical Data Mining”, 2015, IEEE

[3] Yomna M. ElBarawy, Ramadan F. Mohamedt and Neveen I. Ghali,” Improving Social Network Community Detection
Using DBSCAN Algorithm”, 2014, IEEE

[4] Dominik Fisch, Edgar Kalkowski, Bernhard Sick,” Knowledge Fusion for Probabilistic Generative Classifiers with Data
Mining Applications”, 2013, IEEE

[5] Dianwei Han, Ankit Agrawal, Wei−keng Liao, Alok Choudhary,” A novel scalable DBSCAN algorithm with Spark”,
2016 IEEE International Parallel and Distributed Processing Symposium Workshops

[6] Md. Rejaul Karim, and Dewan Md. Farid,” An Adaptive Ensemble Classifier for Mining Complex Noisy Instances in
Data Streams”, 2014, 3rd INTERNATIONAL CONFERENCE ON INFORMATICS, ELECTRONICS & VISION

[7] Karlina Khiyarin Nisa, Hari Agung Andrianto, Rahmah Mardhiyyah,” Hotspot Clustering Using DBSCAN Algorithm
and Shiny Web Framework”, 2014, IEEE

[8] Jie Xu, Kyeong Ho Moon, and Mihaela van der Schaar, “A Machine Learning Approach for Tracking and Predicting
Student Performance in Degree Programs”, 2016, IEEE

[9] S. M. Merchán, and J. A. Duarte, “Analysis of Data Mining Techniques for Constructing a Predictive Model for
Academic Performance”, IEEE Latin America Transactions, Vol. 14, No. 6, June 2016

[10] Ishwank Singh, A Sai Sabitha, Abhay Bansal, “Student Performance Analysis Using Clustering Algorithm”, 2016, IEEE

[11] [Link] Devasia, [Link] T P, [Link] Hegde, “Prediction of Students Performance using Educational Data
Mining”, 2016, IEEE

[12] Yuni Yamasari, Supeno M. S. Nugroho, I N. Sukajaya, Mauridhi H. Purnomo, “Features Extraction to Improve
Performance of Clustering Process on Student Achievement”, 2016, IEEE

[13] Nurul „Ulyani Mohd Najib, Nor‟Aini Yusof, Amin Akhavan Tabassi, “Service Quality Performance of Student Housing:
The Effects on Students‟ Behavioural Intentions”, 2017 IEEE 15th Student Conference on Research and Development
(SCOReD)

© 2018, IJCSMC All Rights Reserved 162