Clustering Data Kecelakaan Lalu Lintas di Indonesia Menggunakan Metode K-Medoids

Abstract

Traffic accidents pose a serious issue in Indonesia. According to data, Indonesia ranks fifth globally in the number of fatalities caused by traffic accidents. Clustering methods are chosen as an approach to identify patterns within accident data, thereby providing valuable information for mitigation efforts. In this study, accident data were sourced from the Indonesian Central Bureau of Statistics, and data pre-processing was conducted to ensure the accuracy of analysis. The K-Medoids algorithm was implemented to partition accident data into several clusters that reflect distinct characteristics of accident-prone areas. This research successfully applied the K-Medoids algorithm to cluster traffic accident data in Indonesia over the 2018 to 2022 period. The analysis results revealed variations in accident patterns across provinces, including differences in the number of cases and severity levels within each cluster. These findings are expected to provide valuable insights to support accident mitigation efforts and the formulation of traffic safety policies in Indonesia.