Implementation of Clustering and Association for Early Warning of Disasters in Bojonegoro Regency

The research aimed to analyze the relationships between different types of disasters, assess the likelihood of disaster occurrences, and enhance knowledge and understanding of disaster patterns in Bojonegoro Regency. The goal was to enable better disaster prediction and preparedness in the future. The methods applied included mapping, clustering using the K-means algorithm, and association rule mining with the Apriori algorithm. Secondary data were obtained from the National Disaster Management Agency and the Bojonegoro Regency Regional Disaster Management Agency Office, covering eight types of disasters. The results reveal that the K-means model groups the data into 5 clusters from 28 sub-districts in Bojonegoro. There are 13 sub-districts in Cluster 0, 1 sub-district in Cluster 1, 4 sub-districts in Cluster 2, 6 sub-districts in Cluster 3, and 4 sub-districts in Cluster 4. The association rule analysis produces four association rules using a minimum support of 10% and a minimum confidence of 50%. The findings highlight that the Ngasem and Bojonegoro sub-districts require more focused disaster management. The fourth association rule has the highest confidence level at 78.79%, indicating that forest and land fires are likely to follow when drought occurs. The research implies that it can support more targeted disaster management focusing on high-risk sub-districts such as Ngasem and Bojonegoro. The originality of the research lies in its novel application of clustering and association rules to analyze disaster patterns in the region, with implications for more targeted disaster mitigation strategies.