<![CDATA[The rapid expansion of language learning in higher education highlights the need for data-driven approaches to monitor student progress and provide timely instructional support. This study aims to develop a predictive framework for Mandarin vocabulary mastery using supervised machine learning. A dataset of 147 undergraduate students was analyzed, incorporating study hours, number of exercises, pre-test scores, and attendance as predictors of learning outcomes. Logistic Regression, Random Forest, and XGBoost algorithms were trained and evaluated, with XGBoost achieving the highest performance (accuracy 88%, F1-score 0.88), demonstrating its superior ability to capture complex learning patterns. Analysis of feature importance revealed that pre-test scores and the number of exercises were the most influential predictors of student success. Furthermore, a prototype graphical user interface (GUI) was developed to visualize predictions in real time, enabling instructors to identify at-risk students and adjust teaching strategies accordingly. The novelty of this study lies in integrating predictive analytics with pedagogical applications, bridging machine learning and educational practice. Beyond its technical contributions, this research provides practical insights for higher education stakeholders, showing how predictive models can support early intervention, enhance curriculum design, and promote evidence-based decision-making in Mandarin vocabulary instruction.]]>
Copyrights © 2026
