Estimating flows between geographical locations: ‘get me started in’ spatial interaction Modelling


In geography and demography, understanding the flows of people, goods or capital between origins and destinations is an enduring challenge. Spatial interaction models have been used for a long time to analyse these flows and to create estimates where data are missing or inaccurate. But implementing spatial interaction models can be a challenge, especially where very few pedagogic guides exist and much of the literature is concerned either with rigorous mathematical derivation of the equations or the results of their application.

This paper is designed as an accessible guide for anyone new to spatial interaction modelling, with the modelling process is detailed using the freely available statistical package, R. Using an empirical migration example it demonstrates how a doubly constrained model can first be fitted to existing flow data with a Poisson regression model to produce parameter estimates. It is then shown how these parameters can be incorporated into an entropy maximising model along with total inflow and outflow data to produce origin/destination flow estimates.

Authors:  Adam Dennett

Publication Date: 28th March 2012

RPS ID:

Download working paper No. 181. File size 816 KB, PDF format.