Comparative Analysis of K-Medoids and Purity K-Medoids Methods for Identifying Accident-Prone Areas in North Aceh Regency

Abstract

Purpose: This study aimed to conduct a comprehensive comparative analysis between K-Medoids and Purity K-Medoids clustering methods for identifying accident-prone areas in North Aceh Regency. The analysis was carried out to provide valuable insights for policymakers and stakeholders to implement targeted interventions as well as improve road safety measures in the region.

Methods: This study compared the performance of K-Medoids and Purity K-Medoids, on accident-prone area data in North Aceh Regency. The algorithm performance was measured using the Davies-Bouldin Index (DBI) method, where a low value signifies superior performance. Additionally, the number of iterations produced by K-Medoids and Purity K-Medoid methods were compared, with lower iterations indicating better performance.

Result: The results showed that Purity K-Medoids had superior performance with an average of 2 iterations and DBI value of 0.7847 across 10 testing runs, while K-Medoids obtained 13.4 iterations and 1.5128, respectively.

Novelty: The study offers valuable insights into the effectiveness and efficiency of clustering methods for identifying accident-prone areas, as guides for policymakers and stakeholders in implementing targeted interventions to improve road safety measures. Additionally, the results provide a methodological framework for evaluating clustering algorithms in similar geographical contexts, enhancing the understanding of their applicability and performance in real-world scenarios.