Improving the Results of Artificial Neural Network Models for Residential Valuation
Peter Rossini · 1998
Abstract: This paper extends work on the application of Artificial Neural Networks (ANN) to residential property valuation which was presented at the PRRES Conference in 1997 where the results of ANN were compared to MRA. In this paper attempts are made to improve the results from ANN Models through the use of more sophisticated methods (such as Genetic Optimization) to derive the appropriate ANN structure. Initial models are created for multiple locations and property types using data supplied by the Department of Environment and Natural Resources in South Australia. Experiments are conducted to see if the inclusion of additional qualitative variables significantly improves the predictive power of the ANN and MRA models. Introduction: The development of Artificial Intelligent Systems for the valuation of residential property is occurring rapidly. In South Australian such a system is under development at the University of South Australia. This development has led to several significant questions, some of which are examined in this paper. In particular