Sistem Cerdas Untuk Meningkatkan Prestasi Siswa Dengan Analisis Rapor Menggunakan Algoritma K-Means Clustering Berbasis Web

Abstract

SMA Negeri 1 Aek Songsongan faces problems in managing and analyzing student academic data, where historical report card data has not been optimally utilized to support appropriate educational decision-making. This study aims to develop a web-based intelligent system using the K-Means Clustering algorithm to automatically analyze patterns of student academic achievement and generate targeted learning recommendations. The research method uses a Mixed Method approach with Knowledge Discovery in Databases stages, including data collection, preprocessing, clustering with K-Means, evaluation using the Davies-Bouldin Index, and provision of intelligent system recommendations. The research data consisted of 1,384 student report card scores from three academic years, which after preprocessing resulted in 1,075 valid data with 8 subject attributes. The clustering results identified three groups of students with a Davies-Bouldin Index score of 0.9678, indicating excellent separation quality. Cluster 0 contained 317 students (29.5%) with high achievement, Cluster 1 contained 706 students (65.7%) with moderate to good performance, and Cluster 2 included 52 students (4.8%) with uneven score patterns. The system generates specific recommendations for each group, ranging from acceleration programs, regular tutoring, to intensive remedial programs, thereby helping schools design more effective and targeted learning strategies.