Comparative Analysis on Educational Data Mining Algorithm to Predict Academic Performance

Abstract

Evaluation of the quality of learning in tertiary institutions is used to determine academic success. The variables used in measuring academic success are “Academic performance, satisfaction, acquisition of skills and competencies, persistence, achievement of learning objectives, and career success. Educational Data Mining (EDM) is a research area that has experienced very rapid development in the last decade. Methods that have been used in evaluating the quality of the learning process. EDM application in tertiary institutions has been the most widely used to predict academic achievement. The methods used in EDM to predict academic achievement include Naïve Bayes classifiers, Support vector machine (SVM), Logistics regression, K-Nearest neighbor (K-NN), ID3 Decision tree, C4.5 Decision tree, Decision tree (DT), Multi-layer perceptron (MLP) neural network, Neural network (NN), Deep learning (DL). Seeing the number of algorithms that have been used In predicting academic achievement, research is needed which can provide recommendations on which algorithm is most appropriate in predicting student academic achievement. This study proposes a method to predict student academic performance by comparing various algorithms used. This study found that the Gradient Boosted Trees algorithm is the most appropriate algorithm for measuring student academic performance.