Document Type



Hedonic regression models are widely used in housing price studies. However, incorrect identification of the functional relationship may lead to potential bias. This research studies the neural network model in modeling housing price. A neural network model is regarded as a nonlinear regression model without a predetermined functional form. The adoption of the neural network model could overcome the biasness inherent in the hedonic price regression model. The performance of neural network models and hedonic regression models are discussed based on the Hong Kong property price data.