Understanding Spatial Dependency Among Spatial Interactions

Spatial dependency exhibits special regularities in spatial interactions. Measuring spatial dependency among spatial interactions can help discover interesting interaction patterns and clusters. Although some metrics have been set up, it is still unclear what potentially affects the presence of spatial dependency among spatial interactions. Thus, we propose an analytical framework to better understand spatial dependency among spatial interactions. First, we define spatial weight matrix for spatial interactions, and then extend Moran’s I and LISA to spatial interactions. Second, we test factors such as first-order spatial autocorrelation and distance decay effect that influence the degree of spatial dependency among spatial interactions. Third, we construct a spatial econometric model for spatial interaction to demonstrate the significance of spatial dependency. The proposed analytical framework is applied in synthetic data and Beijing taxi flows. Results show that the spatial dependency among spatial interactions is positively correlated to the first-order spatial autocorrelation, which is affected by the distance decay effect under a gravity model. Incorporating spatial dependency into a spatial econometric interaction model can also improve its performance.