Project Overview
This studyinvestigates the relationship between street view characteristics and UrbanHeat Island (UHI) effects in New York City by applying machine learningtechniques to street view imagery and socio-demographic data. Using the K-Meansclustering algorithm, four distinct urban clusters were identified: ‘LeafyRetreat’, ‘Suburban Sprawl’, ‘Concrete Jungle’, and ‘Open Horizon’. Theresearch further explores correlations between street views and socio-economicfactors, revealing that tree view is positively correlated with income andnegatively correlated with Black population density, highlightingsocio-economic disparities in access to green spaces. The ‘Suburban Sprawl’cluster has the highest Land Surface Temperature (LST), while the ‘ConcreteJungle’ cluster has the highest Air Temperature (AT) but lower LST. The RandomForest model explains up to 30% of the variation in UHI effects, with tree viewbeing the most important factor.
