Title: A data analytic approach to predicting students' performance using WEKA

Authors: Benjamin Eli Agbesi; Oliver Kufuor Boansi; Prince Clement Addo

Addresses: Faculty of Applied Sciences and Mathematics Education, Akenten Appiah-Menkah University of Skills Training and Entrepreneurial Development, Kumasi, 00233, Ghana ' Faculty of Applied Sciences and Mathematics Education, Akenten Appiah-Menkah University of Skills Training and Entrepreneurial Development, Kumasi, 00233, Ghana ' Faculty of Applied Sciences and Mathematics Education, Akenten Appiah-Menkah University of Skills Training and Entrepreneurial Development, Kumasi, 00233, Ghana

Abstract: Data and record keeping are on the rise. Every day a vast amount of data is kept by different institutions. Educational institutions also keep the educational data of students from the beginning to the end of their studies. Educational Data Mining spotlights on using data mining techniques on educational datasets, with the objective of finding patterns in them to aid the teaching and learning fraternity. This work is a case study in Peki North Circuit schools in the Volta Region of Ghana. It focuses on using the records of Junior High students to compare the performance of five common data mining classification algorithms, Naïve Bayes, BayesNet, IBk, Random Forest, and J48 classifiers, and use the best performing algorithm to determine which factors best affect students' performance. The IBk classifier performed best, giving an accuracy of 89.19%, and indicated that learners' attendance and parental occupation had a minimal impact on their performance.

Keywords: educational data mining; WEKA; text classification; learning analytics; data analytics.

DOI: 10.1504/IJIIE.2023.136176

International Journal of Innovation in Education, 2023 Vol.8 No.2/3/4, pp.196 - 209

Received: 06 Feb 2023
Accepted: 09 Nov 2023
Published online: 19 Jan 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article