DIRECT INDUCTION OF COMPACT RULE-BASED CLASSIFIERS FOR RESOURCE MAPPING

Rule-based classifiers are used regularly with geographical informatio n systems to map categorical attributes on the basis of a set of numer ic or unordered categorical attributes. Although a variety of methods exist for inducing rule-based classifiers from training data, these te nd to produce large numbers of rules when the data has noise. This pap er describes a method for inducing compact rule-sets whose classificat ion accuracy can, at least in some domains, compare favourably with th at achieved by larger less succinct rule-sets produced by alternative methods. One rule is induced for each output class. The condition list for this rule represents a box in n-dimensional attribute space, form ed by intersecting conditions which exclude other classes. Despite thi s simplicity, the classifier performed well in the test application pr ediction of soil classes in the Port Hills, New Zealand, on the basis of regolith type and topographic attributes obtained from a digital te rrain model.