Page 64 - The-5th-MCAIT2021-eProceeding

A Student Performance Model Towards Student Performance

Prediction

a
b
Nor Samsiah Sani , Ahmad Fikri Mohamed Nafuri *
a,b Center for Artificial Intelligence Technology, Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia, 43600
Bangi, Selangor, Malaysia.
*Email: p102926@siswa.ukm.edu.my

Abstract

Student’s performance prediction research on higher education commonly usesexamination grades and results as the main
attribute.Contrastingly,in our research, the aim is todevelopa predictive model ofstudent’s performance using student’s data
that has been enriched with data derived from students activities and employment status as the prediction label.This paper
focuses on the early part of model development which is data preprocessing. In data preprocessing, several measures have
been taken such as data integration, data cleaning, data transformation and correlation analysis using Spearman correlation.
The results show that there are 15significant attributes with higher relation to the employment status.

Keywords: student performance; pre processing; prediction

1. Introduction

The ability to predict student’sacademic performance is very important in the field of education. The impact
on academic performance comes from a variety of sources such as personal, social, psychological and
environmental factors (Al-Hagery et al., 2020). Student’s academic performance issue is highly debated,
especially in tertiary education, because it has a direct impact on employability chances (Bhagavan et al., 2020).
All parties in the field of education such as educational institutions, instructors, students and researchers can
benefit by applying learning analytics on higher education data. For example, Avella et al. (2016)emphasised
that the benefits includetargeted course offerings, curriculum development, student learning outcomes,
personalized learning, improved instructor performance,post-educational employment opportunities and
enhanced research in the field of education. Moreover, the results obtained such as significantly interesting
samples, trends and even hidden information can help stakeholders in improving the process of teaching,
exploration and description of phenomena occurring in the field of education (Osmanbegović & Suljić, 2012).
Therefore, the use of data mining methods in achieving the objectives of study is viewedas very appropriate
because of its ability to handle large amount of data while identifyinghidden patterns and relationships
(Bhardwaj, 2011). Hellas et al. (2018) stated in the study of predicting student performance that some methods
that are often used by researchers can be categorized into several groups, namely classification (supervised
learning), clustering (unsupervised learning), mining (finding frequent patterns and/or extraction of features)
and statistics (correlation, regression and t-test). However, Zulkifli et al. (2019) suggested that predictive
modelling for educational data in Malaysia are still lacking in terms of research number in order to yield a clear
picture on student’s academic performance in academic institutions.
Impact of different attributes on the student performance is widely reviewed and discussed in various
researches. Researchers have included correlation method in analyzing the influencing factors in most of their
works. Hutagaol & Suharjito (2019) studied the correlation between demographic and academic performance
to predict student dropout and concluded that variables such as student’s attendance, homework-grade, mid-test
grade, and finals-test grade, total credit, GPA, student's area, parent’s income, parent’s education level, gender

E- Proceedings of The 5th International Multi-Conference on Artificial Intelligence Technology (MCAIT 2021) [51]
Artificial Intelligence in the 4th Industrial Revolution