Using induction of decision trees to predict greater glider density

Abstract

Evaluates machine induction of decision trees to predict density of greater glider Petauroides volans in forests of SE New South Wales, Australia. Three different algorithms for the induction of decision trees, CN2, 1D3 and CART, were compared with three other models: a linear statistical model, a decision table developed by principal components analysis, and an expert system developed by knowledge acqusition. The machine-learning assisted methods were quicker than knowledge acquisition, gave better predictions that other models, and produced concise instructions for discriminating areas of high and low wildlife density. -from Authors Show more