The purpose of this paper is to apply a direct demand modelling approach for the city of Zurich. The estimated OLS, SAR and GWR models allow identifying influential variables in the demand and also capturing directionality trends. These insights can be useful to complement more traditional approaches by making fairly accurate demand forecasts with less costs and computational requirements. The results show that considering spatial dependency in the models leads to an improved fit and predictive accuracy. However, the failure in capturing spatial autocorrelation with the tested weight matrices suggests that the results must be interpreted with care and confirmed by employing a network-based spatial matrix.