Exploring local well-being and vulnerability through OpenStreetMap: the case of Italy

This article investigates the potential of OpenStreetMap (OSM) data in predicting local well-being and resilience in Italy. The linear Least Absolute Shrinkage and Selection Operator (LASSO) is used to handle multicollinearity problems and select the most influential OSM features. The data-driven approach provides evidence that OSM information is highly correlated with several socioeconomic metrics at a provincial scale (NUTS-3 level). Moreover, it claims that some specific points of interest—e.g., bookmakers—can be used for a rapid territorial appraisal of vulnerable territories, i.e., areas that are affected by economic backwardness, poor institutions, low human capital and that, for these adverse conditions, deserve special attention by policymakers concerned with a reduction of regional disparities. While OSM can become a powerful source for policy planning, monitoring and evaluation, future works in the field should explore the scalability of the approach, its use for forecasting purposes, and the adoption of various models and tools such as machine learning techniques to grasp even non-linear relationships between variables.