<![CDATA[Identification of student academic potential, which still relies on traditional methods, is often inaccurate and time-consuming, potentially hindering early intervention for students who need support. This study offers a solution, comparing the effectiveness of the K-Nearest Neighbor (KNN) and Decision Tree algorithms in classifying the academic potential of 7th-grade students at SMP Negeri 1 Mejobo Kudus and SMP Negeri 2 Mejobo Kudus. This study utilized a dataset of 1,100 student data, with key features including Indonesian and Mathematics scores, reading, writing, and arithmetic test results, and behavioral records. Our goal is to help schools precisely identify students who require special attention early on. This system was developed through comprehensive data collection and the application of refined classification models. The results showed that the KNN model achieved 99% accuracy, while the Decision Tree model fell slightly short at 98%. Despite the high accuracy achieved, cross-validation and in-depth analysis were conducted to ensure model generalization and mitigate potential overfitting. Both algorithms proved highly effective in providing accurate mapping of academic potential, with KNN demonstrating slightly superior performance. With the presence of this web-based system, it is hoped that schools can more easily and quickly identify student potential, reduce misidentification, and make more appropriate and inclusive educational decisions, for the sake of better student learning quality. ]]>
Copyrights © 2025
