<![CDATA[Predicting student dropout risk is crucial for supporting early intervention and accountable academic decision-making. This study proposes a multi-class classification (Dropout, Enrolled, Graduate) using voting (Naïve Bayes and Decision Tree) and Explainable AI to enhance transparency. The dataset consists of 4,424 records with 36 features. Evaluation was conducted using k-fold stratified cross-validation (k=10) and the F1-macro metric. The results show that model performance is relatively close and stable at k=10, so model selection must consider the trade-off between performance and interpretability. The main contribution of this research is a web-based early warning DSS prototype that integrates Voting (NB+DT) with an XAI module (SHAP–LIME) so that predictions can be explained, audited, and followed up with academic intervention recommendations.]]>
Copyrights © 2026
