Identifying at-Risk Students in Educational Metaverse Using TAM and Data Analysis

This study explores the key factors influencing students’ intention to adopt metaverse-based learning platforms by extending the Technology Acceptance Model (TAM). Using survey data from 314 students, a Partial Least Squares Structural Equation Modeling (PLS-SEM) analysis was conducted. The results reveal that Attitude (ATT) is the strongest predictor of Behavioral Intention (BI), while Self-Efficacy for System Use (SSE) shows no significant direct effect, suggesting an indirect influence through mediating variables. To identify students at risk of non-adoption, a k-means clustering approach was applied using BI, ATT, and SSE. A risk group emerged with low scores across all three variables. A real-time risk simulation using synthetic student profiles further demonstrated the practical use of predictive modeling to flag students with low adoption potential. These findings highlight the value of incorporating psychological factors into metaverse learning design and offer practical guidance for early intervention. This work contributes to early risk detection in immersive learning by combining TAM with predictive analytics.