MAPPING DISASTER-PRONE AREAS ON JAVA ISLAND USING THE K-PROTOTYPES ALGORITHM

Clustering in disaster areas is often implemented as a disaster mitigation effort with the aim of minimizing risk. Determining the appropriate clustering method based on the data set will influence the clustering results. K-Prototypes is a clustering method that is capable of handling mixed data, numerical and categorical data, so this method is suitable to clustering disaster prone area with mixed data of disaster factors such as incident intensity, type of disaster, population density, and level of infrastructure vulnerability. This research focuses on disaster prone areas on Java Island and clustering using K-Prototypes to group and map areas that have the highest to lowest levels of disaster vulnerability based on the number of incidents, number of victims, and the amount of damage to facilities and the type of disaster. The clustering results obtained mapping of cities in the province into cluster groups based on the level of vulnerability and calculated potential losses based on disasters in each province. Afterward, the clustering results are used to determine priority areas for disaster mitigation to minimize losses.