Klasifikasi Penerimaan Peserta Didik Baru Berdasarkan Sistem Zonasi Menggunakan Algoritma K-Nearest Neighbors

Syah Fiqri, Muhammad
Andi Taufik

Journal of Informatics, Electrical and Electronics Engineering
Vol. 5 No. 1 (2025): September 2025

The implementation of the zoning system in student admission (PPDB) often raises challenges in determining eligibility based on domicile, requiring a data-driven approach to support the selection process. This study aims to classify new students of SMPN 16 Bogor for the 2025 academic year using the K-Nearest Neighbors (KNN) algorithm. The dataset consists of 1,153 student records with attributes including longitude, latitude, distance from home to school, and zoning labels. Preprocessing involved data cleaning, label encoding, and feature standardization before splitting the data into 75% training and 25% testing sets. The optimal parameter was found at K=18 with a minimum error rate of 0.1591. Experimental results showed an accuracy of 97% for training data and 84% for testing data, indicating that the model performs reasonably well despite signs of overfitting. This research contributes by demonstrating that spatial attributes can be effectively integrated into zoning-based classification and provides a foundation for developing more objective and adaptive decision support systems in the context of student admissions.