Improving Hedonic Office Rent Prediction Models

Modelling and predicting office rent have become one of the major research areas in the real estate field, for both academics and practitioners. The hedonic office rent prediction approach is the most common method based on multiple regression. The major deficiencies of the hedonic model are the multicollinearity problem that may exist between a large number of independent variables, the failure to incorporate the whole range of parameters that may affect the office rent and the interaction effects of the independent variables. The hedonic approach can be improved in two successive steps. First, related parameters can be grouped into factors and their influence, rather than the individual influence of the parameters, can be assessed. In this regard, the principal component approach and factor analysis have been employed. Second, the interaction of the factors can be taken into account by means of factorial design approach. The paper underpins the application of the statistical techniques, namely the principal component analysis, and the factorial design approach, to further improve usability of the hedonic approach.