Predicting building age from urban form at large scale

To stay within 1.5 °C of global warming, reducing energy-related emissions in the building sector is essential. Rather than generic climate recommendations, this requires tailored, low-carbon urban planning solutions and spatially explicit methods that can inform policy measures at urban, street scale. Here, we propose a scalable method able to predict age information different European countries using only open morphology data. We find cross-validated regression models are sufficiently robust generalize unseen cities with mean absolute error (MAE) between 15.3 years (Netherlands) 19.9 (Spain). Our experiments show large-scale improve generalization for predicting across cities, but not needed infer missing data known cities. Filling gaps possible MAE 9.6 16.7 Overall, our results demonstrate feasibility generating contexts Europe informing mitigation policies such as energy retrofits. For French residential stock, predictions target retrofit efforts increase savings by more 50% compared Finally, highlight challenges posed inconsistencies form differences need be addressed an actual roll-out methods.