Forecasting house rental levels: Analytical rent model versus neural network

This paper addresses the need of a house rental model in Townsville, Australia. It then describes two models developed for predicting house rental levels. The first model is analytically built upon a priori selected variables and parameters of rental levels. The second model uses an artificial neural network (ANN) as the basis for rent prediction. Issues in model development and evaluation are discussed. A comparison of the two models indicates that the analytical model performs better than the neural network model.