Neurocomputing Approach to Residential Property Valuation

Abstract

This article introduces neurocomputing—the application of neural networks to problems in business. It describes important characteristics of problems appropriate for neurocomputing. An application to market valuation of residential properties is presented to elucidate neurocomputing project development. The back propagation network paradigm in a PC package is used with training and testing sets selected from data of properties sold in a Midwestern city. A description of the training and testing of the network is presented together with an assessment of the network’s performance. Advantages and disadvantages of the neurocomputing approach are discussed.