(487)EVALUATION OF PREDICTIVE DATA MINING ALGORITHMS IN STUDENT ACADEMIC PERFORMANCE

摘要

Based on the analysis of different metrics, this research identifies the most performing predictivealgorithms in educational data environment using the Faculty Support System (FSS)model, andprovides a summary of current practice and guidance on how to evaluate educational models. Thestudy uses Naive Bayes, Multilayer Perceptron, Random Forests and J48 decision tree induction tobuild predictive data mining models on 111 instances of students' data. We applied 10-fold crossvalidation,percentage split and training set methods on data and performance metrics were used toevaluate the baseline predictive performance of the classifiers. The comparative analysis indicationsthat the Multilayer Perceptron performed best with accuracy of 82% and Random Forests came outsecond with accuracy of 79%, J48 and Na?ve Bayes came out the worst with accuracy of around 60%.The evaluation of these classifiers on educational datasets, gave an insight into how different datamining algorithms predict student performance and enhance student retention.