Classification model for hotspot occurrences using a decision tree method

Abstract

Forest fires in Indonesia mostly occur because of errors or bad intentions. This work dcmon~trates the application of a decision tree algorithm, namely the C4.5 algorithm. lo develop a classification model from forest fire dala in the Rokan Hilir district, Indonesia. The classificalion model used is a collection of lf-TllEN rules that can be used lo predict hotspot occurrences for forest fires. The spatial data consist of the location of hotspot occurrences and human activity factors including the location of city centres, road and river networks as well as land cover types. The results were a decision Lree containing 18 leaves and 26 nodes with an accuracy of 63. 17%. Each leaf node holds po:.itive and negati"e examples of hotspot occurrences whereas the rool and internal nodes contain attribute test conditions: the distance from the location of examples to the nearest road. ri\er. city centre and the land cover types for the area where the c:wmplcs are located. Positive exaniples arc hotspot location:. in the study area and negative arc randomly generated points within the aren at least I km away from any positive example. The clas:.ification model categorized whether the region was susceptible to hohpots occurrences or not. The model can be used to predicl hotspot occurrences in new locations for fire prediction.