Tracing Student Learning Outcome at Historically Black Colleges and Universities via Deep Knowledge Tracing

Abstract

Historically Black Colleges and Universities (HBCUs) in the United States play a vital role in higher education, providing educational opportunities and training, particularly for African American students. Deep Knowledge Tracing (DKT) has emerged as an advanced approach to enhancing higher education outcomes by enabling personalized learning experiences and more accurate assessments of student knowledge mastery. This study is the first to leverage DKT techniques for student learning outcome (SLO) tracing at an HBCU, focusing on Science, Technology, Engineering, and Mathematics (STEM) education at Prairie View A&M University (PVAMU). To systematically validate DKT-based SLO tracing, we construct a comprehensive dataset and employ multiple state-of-the-art (SOTA) DKT models to evaluate SLO tracing performance. The dataset consists of 352,148 learning records from $17,181$ undergraduate students across eight colleges at PVAMU. Experimental results demonstrate the effectiveness of DKT models in effectively predicting SLO. Specifically, Dynamic Key-Value Memory Network (DKVMN) outperforms other models due to its advanced mechanisms for capturing student learning patterns. These findings highlight the potential of DKT in identifying students at risk of underperforming on SLO, enabling proactive interventions to support academic progress and success at HBCUs.

Keywords

• Student learning outcome tracing
• knowledge tracing
• historically black colleges and universities
• STEM education