Sistem Rekomendasi Program Studi berbasis K-Means Clustering menggunakan Asesmen Komprehensif dan Nilai Akademik

Abstract

The high level of inaccuracy in choosing study programmes in Indonesia demonstrates the need for an assessment tool that can map students' potential holistically. This study aims to develop a Dream Major Recommendation System (SRJI) based on K-Means Clustering that integrates six semesters of academic grades and four non-academic assessments to help students at SMAN 3 Batang Hari choose study programmes according to their multidimensional potential. This study uses an applied quantitative design with a Prototype Model through the stages of needs identification, K-Means prototype development, and iterative testing. Data from 152 students were processed using median imputation, Isolation Forest, normalization, and Principal Component Analysis (PCA), resulting in 123 valid datasets clustered using Euclidean distance. The study resulted in a study program recommendation system featuring three optimal clusters with a Silhouette Score of 0.6164 and a Davies–Bouldin Index (DBI) of 0.5684. The analysis identified Cluster 0 (n=42, Social Science 64%, mean score 75.6) with a social-enterprising pattern, Cluster 1 (n=58, Natural Science 72%, mean score 87.5) as an investigative-realistic type, and Cluster 2 (n=35, balanced distribution, mean score 81.2) with realistic-conventional characteristics. Testing results demonstrated an 80% recommendation suitability rate based on validation by Guidance and Counseling teachers. This study contributes an Integrated Multidimensional Student Profiling model that expands the educational data mining literature and supports objective, data-driven decision-making in study program selection.