Spatial machine learning: new opportunities for regional science

This paper is a methodological guide to using machine learning in the spatial context. It provides an overview of the existing spatial toolbox proposed in the literature: unsupervised learning, which deals with clustering of spatial data, and supervised learning, which displaces classical spatial econometrics. It shows the potential of using this developing methodology, as well as its pitfalls. It catalogues and comments on the usage of spatial clustering methods (for locations and values, both separately and jointly) for mapping, bootstrapping, cross-validation, GWR modelling and density indicators. It provides details of spatial machine learning models, which are combined with spatial data integration, modelling, model fine-tuning and predictions to deal with spatial autocorrelation and big data. The paper delineates “already available” and “forthcoming” methods and gives inspiration for transplanting modern quantitative methods from other thematic areas to research in regional science.

Multisource Spatial Data Integration for Use Cases Applications

Big opportunities are given by the reuse and integration of data, which, nowadays, are more and more available, thanks to advances in acquisition and modelling technologies and the open data paradigm. Seamlessly integrating data from heterogenous data sources has been an interest of the geospatial community for long time. However, the higher semantic and geometrical complexity pose new challenges which have never been tackled in a comprehensive methodology. Building on the previous theories and studies, this paper proposes an overarching methodology for multisource spatial data integration. Starting from the definition of the use case-based requirements for the integrated data, it proposes a framework to analyse the involved datasets with respect to integrability and suggests actions to harmonise them towards the destination model. The overall workflow is explained, including the data merging phase and validation. The methodology is tested and exemplified on a case study. Considering the specific data sets’ features and parameters, this approach will allow the development of consistent, well documented and inclusive data integration workflows, for the sake of use cases processes automation and the production of Interoperable and Reusable data.

Creation of a Multimodal Urban Transportation Network through Spatial Data Integration from Authoritative and Crowdsourced Data

One of the most significant challenges in cities concerns urban mobility. Urban mobility involves the use of different modes of transport, which can be individual or collective, and different organizations can produce their respective datasets that, usually, are used isolated from each other. The lack of an integrated view of the entire multimodal urban transportation network (MUTN) brings difficulties to citizens and urban planning. However, obtaining reliable and up-to-date spatial data is not an easy task. To address this problem, we propose a framework for creating a multimodal urban transportation network by integrating spatial data from heterogeneous sources. The framework standardizes the representation of different datasets through a common conceptual model for spatial data (schema matching), uses topological, geometric, and semantic information to find matches among objects from different datasets (data matching), and consolidated them into a single representation using data fusion techniques in a complementary, redundant and cooperative way. Spatial data integration makes it possible to use reliable data from official sources (possibly outdated and expensive to produce) and crowdsourced data (continuously updated and low cost to use). To evaluate the framework, a MUTN for the Brazilian city of Belo Horizonte was built integrating authoritative and crowdsourced data (OpenStreetMap, Foursquare, Facebook Places, Google Places, and Yelp), and then it was used to compute routes among eighty locations using four transportation possibilities: walk, drive, transit, and drive–walk. The time and distance of each route were compared against their equivalent from Google Maps, and the results point to a great potential for using the framework in urban computing applications that require an integrated view of the entire multimodal urban transportation network.