<![CDATA[The high rate of student dropouts is one of the important challenges in higher education because it can affect academic quality, learning effectiveness, and the performance of educational institutions. This condition encourages the need for a prediction system that is able to identify students at risk of dropouts early so that preventive measures can be taken appropriately. This study aims to compare the performance of Logistic Regression and Random Forest algorithms in predicting student dropout based on multi-source. The dataset consists of 4,424 student data with 34 attributes covering academic, demographic, socioeconomic, and academic administration aspects. The research stages include data preprocessing, target transformation into binary classification, feature scaling, data sharing using an 80:20 scheme, and handling class imbalances using the Synthetic Minority Oversampling Technique (SMOTE). Furthermore, a modeling process was carried out using Logistic Regression and Random Forest algorithms to predict the risk of student dropout. Model evaluation was carried out using accuracy, precision, recall, F1-score, and Area Under Curve Receiver Operating Characteristic (AUC-ROC). The results showed that Random Forest performed better than Logistic Regression with an accuracy of 0.884, precision of 0.842, recall of 0.785, F1-score of 0.812, and AUC-ROC of 0.930. Meanwhile, Logistic Regression obtained an accuracy of 0.871, precision of 0.780, recall of 0.835, F1-score of 0.806, and AUC-ROC of 0.928. These results show that Random Forest is more effective in handling complex relationships in multi-source data for student dropout predictions]]>
Copyrights © 2026
