Abstract
Dempster-Shafer theory, an extension of Bayesian statistics, allows for the development of probabilistic belief statements of a hypothesis based on the combination of evidence from varied and irregularly sampled data. It allows for the inclusion of ignorance in developing probabilistic belief statements about occurrences. Ignorance is defined as any area where one cannot postulate a belief, but can not automatically classify the disbelief in the occurrence as (1-Belief). This flexibility in modeling permits us to more objectively apply information where there is uncertainty in the data or in knowledge about that data. This preliminary study for predicting the presence of winter wren in the Harvard Forest, explores the interaction of two belief functions based on species sampling, and fuzzy classification of distance from water.
Original work is by Amina Johnson, The Idrisi Project, Clark University, 950 Main St., Worcester, MA 01610-1477 USA