A Penerapan Data Mining K-Means untuk Penentuan Prioritas Perbaikan Jalan di Kota Pematangsiantar

Abstract

Road infrastructure development plays a vital role in improving community mobility and regional economic growth. However, limited budgets often make it difficult for local governments to determine which areas should be prioritized for road repair. This study aims to determine road repair priorities in Pematangsiantar City using a data mining approach through the K-Means Clustering algorithm. The variables used include road condition percentages (good, moderate, lightly damaged, severely damaged), population size, and population density across eight districts. The data were obtained from the Department of Public Works and Spatial Planning and the Central Bureau of Statistics of Pematangsiantar City in 2023. The evaluation using the Elbow Method, Silhouette Score, and Calinski-Harabasz Score indicated that the optimal number of clusters was three (k=3). The Silhouette Score of 0.3301 and Davies-Bouldin Index of 0.7243 show that the clustering performance is acceptable. Based on the clustering results, three priority categories were identified: (1) high priority—Siantar Utara and Siantar Barat Districts, (2) medium priority—Siantar Timur and Siantar Selatan Districts, and (3) low priority—Martoba, Sitalasari, Marihat, and Marimbun Districts. This research provides data-driven recommendations for local governments to determine effective and efficient strategies for road maintenance and infrastructure budget allocation.