OSM-based Economic Diversity Analysis for Europe's Functional Rural Areas

This dataset and code accompany the research task titled “OSM-based Economic Diversity Analysis for Europe's Functional Rural Areas”, which is part of Sub-task 3.3.3 – Web scraping and nowcasting within the GRANULAR project. The study investigates whether OpenStreetMap (OSM) data can be effectively used to estimate economic diversity in Europe’s Functional Rural Areas (FRAs), where conventional administrative statistics often fall short due to spatial dispersion, timeliness constraints, and functional relevance requirements for targeted rural policy analysis. The dataset includes an estimate of economic diversify for Europe's FRAs based on: A methodological pipeline that transforms OSM points of interest into standardized economic activity categories and estimates business diversity indicators. This includes large-scale data extraction, semantic filtering, classification using sentence-embedding models and generative language models, bootstrap resampling for uncertainty estimation, and predictive modeling to correct for uneven OSM coverage. Validation against Eurostat Structural Business Statistics at NUTS 2 level, demonstrating the ability of OSM-derived indicators to reproduce key aspects of official business concentration patterns, particularly when estimating the Herfindahl-Hirschman Index (HHI). High-resolution estimates of business concentration across more than 4,000 model configurations applied to Functional Rural Areas, offering a more granular view of rural economic structures compared to NUTS 2 level. The accompanying code implements the described methodology and includes: Data processing functions for OSM extraction, filtering, classification, and uncertainty estimation using language models. Predictive modeling functions utilizing generalized additive models (GAMs) for business concentration estimation. Performance evaluation scripts to assess model effectiveness through metrics such as R² and RMSE. Comparative analysis code to identify the most informative OSM categories and understand the impact of national context on prediction quality. The study highlights that, while promising for complementing official statistics in providing functional, place-sensitive insights into rural economic resilience, OSM-derived diversity indicators should not be seen as substitutes for traditional data; instead, they serve to improve spatial granularity, timeliness, and functional relevance where such information is lacking. The scalable, reproducible methodology developed in this research offers valuable potential for rural observatories, territorial monitoring systems, and policy tools aimed at enhancing rural competitiveness and resilience.