<![CDATA[This study addresses the underutilization of student academic data at Universitas Hayam Wuruk Perbanas Surabaya, despite a significant trend of over 50% of students graduating within 3.5 years—raising important questions about academic quality assurance. To respond to this, the study aims to develop a data-driven graduation map by applying the K-Nearest Neighbors (KNN) algorithm to classify students based on their likelihood of graduating on time or late. Using historical academic records from 316 students of the 2022 cohort, the methodology involved data preprocessing, feature selection, implementation in RapidMiner, and algorithm testing. The KNN model was evaluated using a 70:30 training-to-testing split. Results demonstrated strong predictive performance of ROC (Receiver Operating Characteristic), including 95.74% accuracy, 100% precision for predicting delayed graduation, and an AUC (Area Under Curve) of 0.969. While the model was highly effective in identifying on-time graduates, it exhibited slightly lower recall for late graduates. These findings offer practical implications for developing targeted academic support systems and contribute to the broader application of machine learning in higher education analytics in Indonesia. Limitations include the use of data from a single cohort and reliance on one algorithm; future research may explore multi-cohort data and compare multiple classification methods to enhance generalizability and robustness.]]>
