Educational Data Mining to Improve Learner Performance in Gauteng Primary Schools

Educational institutions' main goal is to contribute to the improvement of quality in education. The useful information obtained from the Educational Database Systems can be used in predicting learners' performance. Educational data mining (EDM) is a new emerging data mining (DM) technique used to improve the quality of education. The dataset of learner academic records was applied on Naïve Bayes, BayesNet, JRip and J48 classification algorithms using the Weka tool. The learners were classified based on their city, school, grades, and Mathematics results. About 678 learner's data was covered. This paper showcases the comparison of four classifiers and finds the best performing classification algorithm among all. Based on the results obtained, we found that J48 algorithm outperformed the other algorithms with 99.13% prediction accuracy.