<![CDATA[Student graduation rates are a key performance indicator that’s really crucial for the quality and reputation of a university. However, study delays and the risk of dropping out are still serious challenges for the Informatics Study Program at Dehasen University Bengkulu. This research focuses on tackling these issues by developing an early prediction system for student graduation status to help support more proactive academic decision-making. This research uses an Educational Data Mining (EDM) approach with the Knowledge Discovery in Databases (KDD) framework to process student academic datasets. It implements the K-Nearest Neighbor (KNN) algorithm. Based on the testing results with the hyperparameter K=5, the KNN model achieved an accuracy rate of 50%, which indicates that the decision boundary between students who pass and those who don't cannot be significantly separated by the numeric academic variables currently used. This finding suggests that further development is needed by adding latent variables or sociological data, as well as more specific feature selection techniques to better capture student behavior patterns and improve the model's performance in the future.]]>
Copyrights © 2026
