Klasifikasi Data Titik Api di Bengkalis Riau Menggunakan Algoritme Pohon Keputusan Berbasis Spatial Entropy

Abstract

Forest fire can be monitored using satellite by detecting hotspots as fire indicators at certain times and locations. The purpose of this research is to develop a decision tree for predicting hotspot occurences in Bengkalis district, Riau province Indonesia using the spatial entropy-based decision tree algorithm. The data used in this research are forest fire data of Bengkalis area. The data include city centre, river, road, income source, land cover, population, precipitation, school, temperature, and wind speed. This research, using the 5-fold cross validation, yields five decision trees with the average accuracy of 52.05% and 89.04% on the testing set and the training set respectively.The best accuracy of decision tree is 56% on the testing set that has 560 nodes with the land cover layer as the root node. From the decision tree, as 255 rules for classifying hotspot occurences are obtained. There are 20 objects in the testing set that cannot be classified by the decision tree with the highest accuracy on the testing set.