<![CDATA[While Artificial Intelligence (AI) integrates into higher education to streamline information retrieval, comprehension, and academic task completion, over-reliance on these tools may jeopardize student outcomes. Extant literature predominantly examines AI adoption rates, dependency levels, and user sentiments; however, comparative analyses of machine learning models for classifying academic performance relative to AI usage intensity remain scarce. To address this gap, this study evaluates and compares the efficacy of Naïve Bayes and XGBoost algorithms in predicting student performance based on their AI engagement. Utilizing the Academic Outcomes & AI Dependency Analysis Dataset—comprising 8,000 instances and 26 features—the methodology encompasses data preprocessing, normalization, partitioning, model training, and evaluation via accuracy, precision, recall, and F1-score. The empirical results demonstrate that XGBoost outperforms Naïve Bayes, achieving a superior accuracy of 84.17% compared to 79.03%. Consequently, XGBoost proves to be a more robust model for classifying academic performance driven by AI usage, offering valuable insights for the advancement of educational data analytics.]]>
Copyrights © 2026
