Using machine learning to identify influential factors and predict student academic performance in blended learning

This study aims to identify the factors that influence academic performance and use them to develop a predictive model for student academic achievement, in order to support the improvement of education quality. In previous studies, the selection and evaluation of factors were only conducted on online learning data. In this study, we propose using a selected set of attributes from experimental data collected both in face-to-face classes and on the online learning system at Hanoi National University of Education. To build the predictive model for academic performance, we employed two variable selection methods: one is to choose highly correlated variables, and the other is to use the Stepwise linear regression analysis. Furthermore, two machine learning algorithms, linear regression, and support vector regression were used to construct the predictive model. The experimental results show that the support vector regression model with a polynomial kernel function built from the Stepwise-selected variables is the most effective.