Klasifikasi Kerapatan Titik Api di Bengkalis Riau dengan Menggunakan Algoritme Pohon Keputusan ID3 Spasial

Abstract

Forest fire is considered as a yearly problem in Indonesia. About 20,000 hotspots were recorded each year in the period of 2001-2012 in the Sumatera Island. Hotspot occurrence is an indicator for forest fire events. This research aims to determine the classification model based on historical forest fire data in Bengkalis district, Riau Province Indonesia to predict density of hotspots. The data used in this research are hotspots data in Bengkalis in 2008. Characteristics of the study area includes land cover, income source of community, precipitation, temperature, and wind speed. The method applied in this research is the spatial ID3 decision tree algorithm. This research has successfully produced two decision trees to classify the areas into three categories of hotspot density namely low, medium, and high categories. The highest-accuracy of decision tree is 60.47% on the testing set in the area of Rokan Hilir district, Riau province Indonesia.