A systematic review of regression models for street vitality

Street vitality represents complex phenomena, influenced by multiple interacting factors, but increasingly considered as a mark of successful built environments. Despite the proliferating vitality modelling in urban areas over the past decade, there has been a lack of critical assessment of results’ replicability, particularly concerning whether there are consistent predicting factors and determinants. This systematic review focuses on regression-based studies of street vitality, aiming to comparatively assess the models used, the factors most frequently employed, and their implications. From 62 articles, 155 regression models were extracted, featuring 16 proxies for vitality and 536 significant independent variables. The most robust were the spatial regression models, which successfully predicted 72% of vitality proxies on average. The identified proxies capture the presence of people, user-generated content, street view perceptions, mobility metrics, and other built environment metrics. While the significance of functional diversity and road density as independent variables is reinforced, the relationship between street vitality proxies and other factors is not always consistent, potentially due to contextual and cultural dissimilarities, besides variations in the factors’ measurement or data quality. The emphasis on physical aspects and the high goodness-of-fit observed in certain models may inadvertently diminish the importance of temporal or tactical interventions. Having certain variables serve as both dependent and independent variables in separate models underscores an ambiguity surrounding the vitality notion, while the extent to which different proxies reflect the actual street vitality remains unanswered.