Reflections on the Use of <i>k</i>‐Nearest‐Neighbours Bespoke Neighbourhoods in Urban Studies

Abstract During the first decades of the 21st century, bespoke neighbourhoods have often replaced predefined areas in urban studies. This development has been enabled by the improved availability of detailed data, as well as methodological innovations (e.g. the construction of bespoke neighbourhoods based on k ‐nearest neighbours, or k nn). This paper argues that, while k nn‐based bespoke neighbourhoods facilitate more nuanced results in several respects, they do not provide all the advantages claimed by their advocates; as have been pointed out, they also present new problems. A literature review of the arguments for k nn, illustrated by concrete examples of how k = 400 neighbourhoods fit the varying physical urban environments, leads to the conclusion that, despite explicit claims to the contrary, the modifiable areal unit problem (MAUP) still remains, and certainly various physical aspects of urban space, are largely ignored. Finally, this article discusses various possible ways to include aspects currently missing.