Leveraging ENR-GLM (Elastic-Net Regularized Generalized Linear Model) and Neural Networks for Predicting Student Dropout and Academic Success

• In Event: Educational Research in Transition: From Classical Methods to Machine Learning

Abstract

This study employs elastic-net regularized generalized linear models for feature selection, followed by a comparative analysis of neural networks and logistic regression in predicting student dropout and academic success. The results indicate that neural networks outperform logistic regression for predictive accuracy. This finding highlights the limitations of previous practices that classify students and identify those at-risk with low accuracy. By leveraging the strengths of elastic-net regularization for feature selection and the advanced predictive capabilities of neural networks, this research underscores the potential of machine learning techniques to enhance educational outcomes by more effectively identifying at-risk students. Furthermore, this study contributes to the growing body of evidence supporting the integration of sophisticated analytical methods in educational measurement and student support systems.

Authors

• Xiaoting Zhong, University of Iowa

• Tao Wang, University of Victoria