Models for Spatial Interaction Data: Computation and Interpretation of Accessibility

The paper is in the area of spatial optimization, for simulating and understanding spatial interaction. The model foundation is from Wilson's "doubly constrained" model of spatial interaction. The idea is to perform a sensitivity analysis on the parameters of this model and to interpret the results in terms of accessibility. The main purpose of this type of analysis is to use data from interaction systems to uncover structural effects that help to understand the role of origin and destination location, and accessibility. US air passenger traffic is used as the starting point for the model. The model reproduces many of the features of the data with a parsimonious set of parameters, leaving some aspects of the analysis open to interpretation. An innovative idea in this paper is to compute averages and other measures directly from the data, fit a model to these data, and then to use the fitted (and observed) matrices to evaluate numerous theoretically inspired measurements. This paper (in a modular way) develops the introduction and context, and then moves to theory, spatial disaggregation, and empirical applications.