Information Theory and Classification in Geography

Abstract

<p> In this paper some fundamental concepts of information theory and their potential for classification construction in geography are discussed. The concepts of information and uncertainty are shown to be equivalent. Three different information measures are discussed and the particular situations for which each is appropriate are identified. With this background the available literature on the application of information theory to classification is reviewed and reinterpreted. A classification algorithm for objects characterized by multistate ordinal attributes is presented and tested. Recommendations for further research include the consideration of explicitly spatial information measures and the examination of more general information metrics.</p>