The number of choice alternatives in a spatial context can be very large, often in the thousands and sometimes millions.

Substitution patterns are the impacts on the distribution of choices on alternatives, when the attractiveness of some alternative changes. Substitution patterns are typically very complex and difficult to capture in a controlled and transparent way with current models.

Second, the demand for specific choice alternatives affects their attractiveness through strong feedback mechanisms related to pricing, congestion and various social interactions. This is an endogeneity issue, and it is hard to handle with current models.

Failure to handle endogeneity leads to bias, which may render predictions meaningless. Both these issues need to be addressed simultaneously and the current state of the art does not do this.

This project will develop a new category of spatial choice models, designed to overcome these two serious short comings of the current spatial choice models. The new spatial choice models are based on a generalisation of entropy.