Identifying brown bear habitat by a combined GIS and machine learning method

In this paper we attempt to identify brown bear (Ursus arctos) habitat in s outh-western part of Slovenia, a country lying on the north-western-most ed ge of the continuous Dinaric-Eastern Alps brown bear population. The knowle dge base (in the form of a decision tree) for the expert system for identif ying the suitable habitat, was induced by automated machine learning from r ecorded bear sightings, and then linked to the GIS thematic layers for subs equent habitat/non-habitat classification of the entire study area. The acc uracy of the decision tree classifier was 87% (KHAT 73%). The decision tree mostly agreed with the existing domain knowledge. For the study area the m ain factors considered by the expert system to be important for brown bear habitat were the percentage of forest (positive), proximity to settlements (negative) and elevation above see (positive), however the decision tree di d not account for habitat patch size. After filtering out habitat patches s maller than 5000 ha in GIS, the accuracy increased to 89% (KHAT 77%). Where as 88% of the habitat was within forests, only 33% of all forests were cons idered suitable as habitat. (C) 2000 Elsevier Science B.V. All rights reser ved.