k-means and GIS for Mapping Natural Disaster Prone Areas in Indonesia
Suwardi Annas (a*), Zulkifli Rais (a)

a) Department of Statistics, Universitas Negeri Makassar
Jalan Mallengkeri, Makassar 90224, Indonesia
*suwardi_annas[at]unm.ac.id

Abstract

The number of natural disasters in Indonesia is very high occurrences. However, the data collected based on natural disasters has a complexity data structure. One of the efforts to make prevention by grouping the areas of natural disaster. The proposed methods to analyze the data are k-means and Geographical Information System (GIS). The k-means method has mapped the areas of natural disaster based on districts into 3, 4, 5, 6 and 7 of clusters. This result showed that the best cluster resulted by 7 of clusters with the smallest root mean square standard deviation (RMSD) than other clusters. Although k-means has obtained the best cluster, however, it was difficult to present the clustering of natural disaster districts in the map. Therefore, a GIS method was used to improve the cluster visualization of k-means. The main purpose of GIS is to develop a visual map of the natural disaster districts according to a given cluster of the k-means. GIS method can be a useful tool to improve the visualization information of k-means clustering and enables interpretation of the disparities of natural disaster by districts in Indonesia.

Keywords: GIS; k-means; Natural disaster; RMSD

Topic: Mathematics