<![CDATA[Student absenteeism, particularly unexcused absence, remains a critical challenge in basic education management that negatively impacts academic continuity and increases dropout risks. This study presents a predictive analysis model for student absenteeism levels using two machine learning algorithms: Naïve Bayes and Logistic Regression, applied to 1,533 active student records from SMP Negeri 5 Stabat across the 2021-2025 academic years. Predictor features comprise demographic factors, accessibility metrics, and parent socioeconomic indicators. Automated data processing was executed via a Python API backend connected directly to a MySQL database across five computational stages. Model evaluation was conducted under three train-test split scenarios (70:30, 80:20, and 90:10). Empirical results demonstrate that Logistic Regression consistently outperformed Naïve Bayes across all testing configurations. The highest classification performance was achieved by Logistic Regression under the 90:10 split ratio with an accuracy of 84.42%, while achieving 84.36% accuracy, 0.8421 precision, 0.8436 recall, and an F1-score of 0.8422 under the standard 80:20 split ratio. Conversely, Naïve Bayes yielded inferior generalization due to feature multicollinearity, recording its lowest performance at 62.34% under the 90:10 ratio and 64.17% under the 80:20 ratio. Sigmoid logit transformation in Logistic Regression proved highly robust in handling interdependent socioeconomic and demographic attributes. These findings confirm the efficacy of LR-based Decision Support Systems for early warning intervention in educational institutions.]]>
Copyrights © 2026
