Evaluating various machine learning methods for predicting students' math performance in the 2019 TIMSS

Abstract

The growth of a country strongly depends on the quality of its educational system. All over the world, the education sectors are experiencing a fundamental evolution of their mode of operation. The greatest challenge for education today is the low success rate of learners and the abandonment of education in institutions at a premature age. Early prediction of student failure can help administrators provide timely guidance and supervision to enhance student success and retention. We propose a performance prediction model based on students' social and academic integration using several classification algorithms. This study involves a comparative analysis of five algorithms: logistics regression, k-nearest neighbors (K-NN), support vector machine (SVM), decision tree, and random forest. They were applied to a set of data from TIMSS 2019 in Morocco, to determine their effectiveness in predicting student performance using prediction models such as logistics regression, KNN, SVM, decision-tree, and random forest, decision-makers can make data-driven decisions to enhance educational strategies and improve outcomes in mathematics education.