Exploring the Efficacy of Machine Learning Techniques for Predicting Students’ Academic Performances

Abstract

Predicting students' academic success in tertiary institutions in terms of determining accurately whether a student will be a top 10 in his class, an average student or a dropout is a significant issue in higher education. Therefore, to predict the behavior of a learner, many data mining techniques are used, such as clustering, classification, and regression. In this paper, students’ academic performance prediction model and new features are introduced that have a great influence on student’s overall academic achievement. Twenty features were considered from the dataset obtained from about 200 students using questionnaire and information from the institutions database. The Naive Bayes classifier was used and the experiment was carried out in a WEKA implementation workbench. The attributes were first analyzed for relevance, the analysis showed that a student’s first semester grade point average (GPA) has the highest relevance followed by friends’ study affinity, learning facilities, private home lesson, gender and available transport facilities. One hundred and sixty-nine (169) students’ records were used for the analysis, the model recorded 162 instances that were correctly classified which translates to 95.86% accuracy and incorrectly predicted 4.142%. The experiment showed an excellent performance from the Naive Bayes classifier in predicting the dataset correctly Keywords: Educational Data Mining, Students' Performance Management, Tertiary Institutions, Classification, Naive Bayes Classifier. Agwi, C.U. & Akpojaro, J. (2024): Exploring the Efficacy of Machine Learning Techniques for Predicting Students’ Academic Performances. Journal of Advances in Mathematical & Computational Science. Vol. 12, No. 1. Pp 53-66. Available online at www.isteams.net/mathematics-computationaljournal. dx.doi.org/10.22624/AIMS/MATHS/V12N1P6