|
Abstract: Spatial interaction modelling is one of the major intellectual achievements and at the same time, perhaps the most useful contribution of spatial analysis to social science literature. The interest in such models is motivated by the need to explain flows of tangible entities such as persons and commodities, or flows of intangible entities such as information and knowledge across space.
The models typically rely on a discrete representation of space and three types of variables to explain mean interaction frequencies between origins and destinations of interaction: (i) origin-specific variables that characterize the ability of origins to generate flows, (ii) destination-specific variables that represent the attractiveness of destinations, and (iii) origin-destination-specific variables that characterize the way, spatial separation of origins from destinations constrains or impedes the interaction.
The basic units of analysis are origin-destination [OD] pairs of location. The focus of my talk is on neural spatial interaction models. They are closely related to spatial interaction models of the gravity type and termed neural in the sense that they are based on neurocomputing, inspired by neuroscience, derived from single-hidden layer feedforward neural networks.
|
