A transformer-enhanced framework for estimating student performance in mathematics

Estimating student performance is a crucial focus in educational research, significantly impacting academic outcomes and instructional strategies. This article investigates the application of advanced deep learning techniques for predicting student performance based on heterogeneous tabular data. We propose a novel hybrid architecture that combines a Hybrid Transformer module with a Multi-Layer Perceptron (MLP) layer, designed to effectively capture both static and dynamic feature interactions. Motivated by recent advancements in Transformer-based models for tabular data, our approach integrates self-attention mechanisms with feed-forward pathways to enhance feature learning. Experimental results demonstrate that the proposed model outperforms traditional machine learning methods and state-of-the-art tabular deep learning architectures across multiple evaluation metrics. These findings highlight the potential of hybrid Transformer-MLP architectures in optimizing predictive analytics for educational outcomes.