Data-driven and machine-learning approaches for just and healthy urban planning

Climate change is intensifying the frequency, severity, and duration of extreme weather events such as heatwaves and wildfires, disproportionately affecting vulnerable urban populations, particularly the elderly. This thesis presents a dual risk assessment focused on populations aged 65 and older in Bologna, Italy, and California, the United States. By integrating environmental, infrastructural, and demographic data, it identifies susceptibilities and proposes data-driven mitigation strategies using Machine Learning (ML). Using wildfire (2014–2021) and heatwave (2014–2023) data at the census tract level, the study develops risk maps under Representative Concentration Pathways (RCP) scenarios 4.5, 6.0, and 8.5. It applies Random Forest (RF) and Long Short-Term Memory (LSTM) models to identify high-risk urban areas, advocating for transdisciplinary and equitable urban resilience strategies. The methodology builds upon the risk triangle, hazard, exposure, and vulnerability. Hazards include temperature extremes and wildfire intensity. Exposure is assessed through population density in hazard-prone areas. Vulnerability accounts for age, gender, socioeconomic status, health, and built infrastructure. Advanced Geographic Information Systems (GIS) and ML methods enhance the assessment’s scalability and accuracy over traditional approaches. This dissertation documents and evaluates the methodology through a structured three-phase approach: creation, analysis, and application of risk maps. The creation phase uses ML to model spatial and temporal risk dynamics. Case studies in Bologna and California demonstrate the integration of diverse datasets, including meteorological, topographical, and demographic variables. The analysis phase optimizes model performance using GridSearchCV and the Hippopotamus Optimization Algorithm (HOA), validated through confusion matrices, statistical metrics, and ROC curves. In the application phase, risk maps inform urban planning and highlight strategies to protect the elderly. This study develops a data-driven framework for climate risk assessment and mitigation. By merging Nature-Based Solutions (NBS), ML, and urban planning, it promotes healthier, more equitable, and climate-resilient cities.