A Decision Tree Based on Spatial Relationships for Predicting Hotspots in Peatlands

Abstract

Predicting hotspot occurrence as an indicator of forest and land fires is essential in developing an early warning system for fire prevenUon. This work applied a spatial decision tree algorithm on spatial data of forest fires. The algorithm is the improvement of the conventional decision tree algorithm in which the distance and topological relationships are included to grow up spatial decision trees. Spatial data consisted of a target layer and ten explanatory layers representing physical. weather. socio-economic and peat/and characteristics in the study area Rokan Hilir District. Indonesia. Target ob;ects were hotspots of 2008 and non-hotspot points. The result was a pruned spatial decision tree with 122 leaves and the accuracy of 71.66%. The spatial tree has produced higher accuracy than the non-spatial trees that were created using the 103 and C4.5 algorithm. The 103 decision tree had accuracy of 49.02% while the accuracy of C4.5 decision tree reached 65.24%.