Spatial interaction (SI) is the process whereby entities at different points in physical space make contacts, demand/supply decisions or locational choices. The entities can be individuals or firms and the choices can include housing, jobs, production quantities, exports, imports, face-to-face contacts, schools, retail centres and activity centres. The first SI models can be grouped under the generic heading gravity models. Their main characteristic is that they model the behaviour of demand or supply segments, rather than that of individuals and firms. This article traces the development of these models from their inception in the early part of the twentieth century to the present. The key advances include the replacement of the gravity analogy by the more general concepts of entropy or information theory, a statistical framework commonly used in physics. With the arrival of the regional science paradigm over 50 years ago, a key challenge has been to broaden these models compared to those arising in spatial economics, thus arriving at a more inclusive probabilistic framework. These efforts are discussed here, as well as inclusion of geographical advances, embracing activities as generators of travel, time-geography, recognition of spatial interdependencies, and use of neuro-computing principles. Copyright Springer-Verlag Berlin/Heidelberg 2003Spatial interaction, gravity models, entropy, probabilistic, spatial structure, neural networks,

Properties of spatially explicit data are often ignored or inadequately handled in machine learning for spatial domains of application. At the same time, resources that would identify these properties and investigate their influence and methods to handle them in machine learning applications are lagging behind. In this survey of the literature, we seek to identify and discuss spatial properties of data that influence the performance of machine learning. We review some of the best practices in handling such properties in spatial domains and discuss their advantages and disadvantages. We recognize two broad strands in this literature. In the first, the properties of spatial data are developed in the spatial observation matrix without amending the substance of the learning algorithm; in the other, spatial data properties are handled in the learning algorithm itself. While the latter have been far less explored, we argue that they offer the most promising prospects for the future of spatial machine learning.

This study empirically investigates the causes of urban traffic congestion to bring into focus the variety of beliefs that provide support for policy interventions to mitigate traffic congestion in USt cities. We use a structural equation modeling (SEM) framework and position the analysis at the meso-scale (i.e., neighborhood shapes, sizes, density, land-use mix, street design, distribution of open space) to better align with policy and planning decisions and strategies. The analysis is carried out on 100 metropolitan areas in the USA, with three complementary metrics of urban traffic congestion and 25 factors representing the structural, socioeconomic, and behavioral aspects of urban areas. SEM results demonstrate that congestion is a complicated phenomenon where indirect effects are pathways powerful enough to offset direct effects under certain circumstances. We find that, beyond the role of urban population size, income and employment agglomeration lead to further traffic congestion. In contrast, the most influential tempering effects come from congestion’s own self-regulation impact, non-car mode choice behaviors, adequate highway transportation, focused community structures, urban density, and socioeconomic factors like car ownership. The article discusses the policy implications of this meso-scale empirical analysis.