Feri Hari Utami

The Student Information System (SIS) plays a crucial role in managing educational data, supporting policy-making, planning, and budget allocation, including the School Operational Assistance (BOS) program. However, data anomalies such as duplicate identities, entry errors, and administrative inconsistencies undermine data validity. Manual anomaly handling is inefficient, particularly at a national scale. This study proposes an unsupervised machine learning approach using the One-Class Support Vector Machine (OC-SVM) to automatically and continuously detect anomalies in educational datasets. A total of 25,204 records from Bengkulu Province were used, with preprocessing involving cleaning, encoding, scaling, and data splitting. Model training includes hyperparameter tuning for optimal performance. Evaluation is conducted using accuracy, precision, recall, F1-score, and confusion matrix. Experimental results show that the OC-SVM model successfully detected 14,424 anomalies with an accuracy of 95%. These findings highlight the potential of automated approaches to improve the quality of national educational data and support future integration into SIS dashboards and further development through ensemble methods. Keywords—Student Information System, One-Class SVM, Unsupervised Learning, Deteksi Anomali.