Application of K-Means Algorithm for Clustering Regions Based on Sanitation Level: A Case Study in Bangka Regency, Indonesia

Dengue Hemorrhagic Fever (DHF) cases have increased annually, especially in Bangka Regency, Bangka Belitung Islands Province. Environmental sanitation is one factor that also influences the incidence of DHF. This research aims to cluster regions based on sanitation level by applying the K-Means Clustering method. The optimal number of clusters is determined using the Silhouette and Elbow methods. The results of this study indicate that the optimal number of clusters for grouping regions based on the level of sanitation using the elbow method is 3 clusters with an inertia value of 8623.87632779. Whereas in the Silhouette coefficient method, for the same number of clusters, the Silhouette coefficient value is 0.396250. The results of area clustering based on the level of sanitation using the number of clusters of 3 yielded: 112 regions included in Cluster 1 (low sanitation category), 263 regions included in Cluster 2 (medium sanitation category), and 105 regions included in Cluster 3 (high sanitation category). The system performance evaluation using the K-Means Clustering method obtained accuracy = 0.8645, average sensibility = 0.8775, average specificity = 0.9231, and average precision = 0.9176. This indicates that the K-Means method can recognize patterns well and differentiate sanitation levels in various regions. The use of the system with the K-Means Clustering method can support analysis for the government, especially the Bangka District Health Office, in identifying regions based on their level of sanitation, which can be used as a basis for evaluation and formulation of strategic steps to improve sanitation to control or anticipate DHF incidents.