PERBANDINGAN ALGORITMA K-PROTOTYPES DENGAN AGGLOMERATIVE CLUSTERING DALAM SEGMENTASI SISWA BERDASARKAN FAKTOR AKADEMIK DAN SOSIAL

Muchamad Risqi
, Muhammad Nashif Farid
, Mohammad Idhom
Trimono Trimono

INFORMATIKA
Vol. 5 No. 1 (2025): MEI : JURNAL INFORMATIKA DAN MULTIMEDIA

Student performance is affected by internal and external factors such as study time, absenteeism, tutoring, and parental support—factors often overlooked by traditional education methods. This study applies K-Prototypes and Agglomerative Clustering with Gower Distance to segment students using mixed-type data. Five key variables were analyzed: study time, absences, GPA, tutoring, and parental support. The Elbow Method was used to identify the optimal number of clusters, and Silhouette Score to evaluate performance. Results show K-Prototypes outperformed Agglomerative Clustering (0.332 vs 0.186). Three student segments emerged: active students with average GPA, low-risk learners, and high-achievers with minimal external support. These findings can inform more adaptive and data-driven academic interventions for education stakeholders.