Analysis of Hotspot Data for Drought Clustering Using K-Means Algorithm

Abstract

Drought is a disaster that is often experienced in Indonesia. This disaster occurred because Indonesia's geographical location is on the equator. Drought has had a major impact on the community such as crop failure, forest fires, soil damage, the emergence of disease outbreaks, and the extinction of animals and plants. Based on data from the Ministry of Environment of the Republic of Indonesia, the distribution of Riau's hotspots is quite unique. It is said so, because in this distribution, Riau has increased in every February and March as many as 277 and 248 hotspots in the last two years, namely between 2018 and 2019. To anticipate the drought that occurred in Riau, the clustering of drought-prone areas was conducted based on the analysis of hotspots data. This clustering of vulnerable areas is done by the K-Means algorithm. In determining the number of clusters of vulnerable areas, the elbow method is used as a determinant and produces as many as 4 cluster. The results of these method were analyzed by the silhouette coefficient. The result of analyzed is 0.388632163 and were classified as well-clustered. From these results, Rokan Hilir, Bengkalis, Kota Dumai are the dangerous district with 3106, 2361, and 117 point of dangerous distribution, respectively.