Integrating Artificial Neural Networks and GIS for Single-property Valuation

An Automated Valuation Model (AVM) named GINS is developed as an alternative for use in the valuation of single-residential property and indicates that GINS provides an efficient AVM tool that provides superior residential property valuations, while accuracy is improved by minimizing the influence of subjective judgments.

Abstract

Abstract: The sales comparison approach to valuing a single property combines expert qualitative evaluation with quantitative analysis of the subject property along with comparables. Predicted values are then evaluated before a final valuation is rendered. The traditional hedonic approach involving the application of multiple regression, is problematic and faces some data management problems. This is especially when large dataset is being analysed and the effect of location attribute cannot be explained explicitly. An Automated Valuation Model (AVM) named “Geo-Information Neural System ” (GINS) is developed as an alternative for use in the valuation of single-residential property. This system integrates a Geographic Information System (GIS) technique with Artificial Neural Networks (ANN) modeling. It involves the establishment of a GIS database management system for all detached property in Damansara Heights, Kuala Lumpur. GIS is utilized for location distance measurements, spatial queries and thematic mapping whilst ANN is employed to replicate the way the human brain might process data by learning relationships, in this case the one existing between property characteristics such as physical and location attributes and sales price. A sample of 100 sales of detached houses is taken. The trained neural network model is then used to predict the probable value of a residential property. The model is built on a GIS platform, which will allow for GINS automation as well as the conduct of interactive valuations. A graphical user interface is developed for seamless integration and user interaction. The results indicate that GINS provides an efficient AVM tool that provides superior residential property valuations, while accuracy is improved by minimizing the influence of subjective judgments. The technique may be used to check valuations generated by more traditional methods as well as further improve the overall quality of single-property valuations. [Keywords: single-property valuation; automated valuation model; geographic information system; artificial neural networks]