Mapping the DNA of Urban Neighborhoods: Clustering Longitudinal Sequences of Neighborhood Socioeconomic Change

The spatial pattern of longitudinal trends in neighborhood socioeconomic dynamics has long been implied by traditional urban models dating back to the Chicago School; however, empirical studies beyond the mapping of change between two points in time are surprisingly limited. This article introduces a methodology to the study of spatial–temporal patterns of neighborhood socioeconomic change. The approach first involves establishing discrete classes of neighborhoods following a k-means clustering procedure and then applies a sequential pattern mining algorithm to determine the similarity of longitudinal sequences. Sequences are then clustered to derive a typology of neighborhood trajectories. The method is employed in an empirical analysis of neighborhood change from 1970 to 2010 for all census tracts in the cities of Chicago and Los Angeles. In Chicago, this time period was marked by a sustained process of center city revitalization through two distinct upgrading processes, whereas in Los Angles, neighborhood upgrading largely came in the form of suburban upgrading. The spatial structure of neighborhood dynamics in Chicago resembled patterns described by Chicago School theorists, whereas the dynamics of Los Angeles deviated from this ordered regularity.