Comparison of K-Medoids and Self Organizing Maps Algorithm in Grouping Hydrometeorological Natural Disasters in Java Island

Abstract

Indonesia is a country that is prone to disasters. Based on data from the National Disaster Management Authority, there are 6868 disasters in the last 3 years. Changes in weather, temperature, and wind direction that are quite extreme in Indonesia lead to one of the disasters namely hydrometeorological disasters such as floods, landslides, tornadoes, forest fires, and drought. According to the National Coordinating Agency for Disaster Management, this type of disaster is quite common in Indonesia. Over the past 3 years, Java Island received the highest number of disasters. The National Disaster Management Authority states that tornadoes contribute 21% of all disasters in Indonesia. Therefore, the researcher wants to grouping the cases of hydrometeorological natural disaster in Indonesia, especially in Java. This research is to determine the areas that are the focus of the government in disaster mitigation efforts using K-Medoids Clustering and Self Organizing Maps. Based on the results of this study, it was found that the suitable method used was the Self Organizing Maps Method because it was seen based on a smaller standard deviation value. Cluster one there are 108 members who are groups with relatively low disaster vulnerability, while cluster two there are 10 members are city and district groups that have relatively high disaster vulnerability.