Web-Based Clustering Application Using Shiny Framework and DBSCAN Algorithm for Hotspots Data in Peatland in Sumatra

Abstract

Forest and land fires currently have become serious problems in Indonesia. Peatlands are frequently burnt because of its characteristic, i.e. combustible when it was in dried condition. In the previous work, hotspots as on indicators for forest and land fires including in peatland were analyzed by applying density based clustering algorithm namely Density-Based Spatial Clustering Algorithm with Noise (DBSCAN). Clustering results hotspots distribution that can be used for preventing and controlling fire events. This research aims to build a web-based clustering application for grouping hotspot data in peatlands in Sumatra using the Shiny framework, that is available in programming language R. Clustering was performed on hotspots data in peatland in 2002 and 2013 using the DBSCAN algorithm. This algorithm finds clusters by identifying areas that have a high hotspots density. The application was successfully built and has several features, namely: a) clustering hotspots, b) visualization of clustering results based on a type of landuse, land depth, and peat type, c) providing the value of within cluster for cluster evaluation, and d) displaying a summary of clustering results. These features have been tested using the blackbox approach and the test results show that the features work properly and produce outputs in corresponding to the test scenario.