Neighborhoods are critical arenas where urban form, accessibility, and daily life intersect, yet most megacities rely on coarse administrative boundaries that obscure local spatial and social dynamics. There is a persistent gap in objective, multicriteria, and data-driven methods for defining neighborhoods and supporting local planning. This paper proposes a transferable, data-light framework that clusters urban blocks into contiguous neighborhood units, contributing to a consistent socioterritorial definition. The framework incorporates barrier-aware adjacency constraints, PCA and multi-index/ARI-stability evidence based protocol to enable urban delineation. The method integrates three key planning dimensions: (A) built environment, (B) accessibility, and (C) sociodemographic context. After standardization and dimensionality reduction via Principal PCA, clustering is performed using three different algorithms. The algorithm dynamically adjusts the k-optimal value for each case. Applied to São Paulo, a 11.4-million–inhabitant city in Brazil, results indicate distance to high-capacity public transit as the most influential factor, correlating strongly with land value and commercial-service concentration. Population density remains relevant but not deterministic, underscoring the importance of a multi-criteria approach to neighborhood analysis. The clustering reveals socio-spatially cohesive areas that cut across formal administrative boundaries, exposing neighborhood-scale structures often obscured in conventional planning units. Core transit-rich clusters concentrate up to 75% of built area in commercial or service use, while peripheral zones remain underserved. The framework identifies neighborhood areas of influence that can support the delineation of reference perimeters for guiding the public policies, and offers planners a flexible tool for neighborhood-scale policy design, inclusive urban governance, and equitable spatial interventions. • Delimits objectively neighborhoods through unsupervised learning method • Replicable clustering approach strengthens neighborhood delineation in cities • Multi-metric clustering and validation method to analyse urban dynamics • São Paulo application evidences clustering’s relevance to neighborhood planning • Results showed strong alignment between cluster patterns and transit accessibility
Aquino et al. (Sun,) studied this question.