The complexity of urban form’s impacts on residents’ neighborhood perception: Machine learning evidence from Singapore

How neighborhoods are defined has long been a key issue in urban research. Traditional approaches primarily rely on census data. However, such demarcations have been increasingly criticized for failing to capture the area that residents perceive as their neighborhoods. These ‘perceived neighborhoods’ more accurately align with their lived experiences. Therefore, identifying perceived neighborhoods would be helpful for understanding residents’ everyday activities and addressing their needs through planning and design. Recently, emerging evidence suggested that residents’ neighborhood perception is significantly influenced by urban form. However, existing studies often oversimplify the influence, tending to examine a single aspect of urban form independently but neglecting the interplay of different aspects, and produce partial understandings. Drawing on diverse urban datasets and cognitive mapping surveys from Singapore, this study develops a machine learning framework to bridge this gap. Using SHAP analysis, the results indicate that residents’ neighborhood perception is primarily affected by urban form features associated with their daily activities—such as footpath density and commercial facility distribution—rather than visual ones. Moreover, how residents perceive a place is affected not only by features of that place itself, but also by those of their own places of residence. It suggests that neighborhood perception is a relative process. Additionally, the findings uncover nonlinear effects of certain features, wherein the effects can diminish or even reverse when over-supplied. This study offers an in-depth understanding of the complexity of urban form’s influence on residents’ neighborhood perception and provides insights for responsive neighborhood planning and design. • Unpack complex effects of urban form on neighborhood perception using XGBoost. • Urban form features related to residents’ everyday lives are more influential. • Neighborhood perception is jointly influenced by urban form features of both residents’ living places and surrounding areas. • Some urban form features exhibit nonlinear effects with oversupply imposing adverse impact.