Indonesia is one of the countries with a high level of vulnerability to natural disasters, making accurate risk mapping essential to support mitigation planning. This study aims to cluster the provinces of Indonesia based on disaster occurrence characteristics using a hybrid approach of Self-Organizing Maps (SOM) and K-Means. The data were obtained from the Indonesian National Disaster Management Agency (BNPB), covering the frequency and types of disasters such as floods, extreme weather, eruptions, abrasion, earthquakes, forest/land fires, droughts, and landslides. The SOM representation results were clustered using K-Means, with the optimal number of clusters determined through the evaluation of the Davies–Bouldin index, Silhouette coefficient, and connectivity measure. The analysis revealed that two clusters provided the best separation: Cluster 1 includes most provinces with medium to low multi-hazard risk, while Cluster 2 consists of West Java, Central Java, and East Java, which have high hydrometeorological risk. This hybrid SOM and K-Means approach successfully identifies the spatial patterns of disaster risk and can serve as a reference for the government in formulating region-based mitigation strategies.