Identifying Neighbourhood Change Using a Data Primitive Approach: the Example of Gentrification

Abstract

Data primitives are the fundamental measurements or variables that capture the process under investigation. In this study annual data for small areas were collated and used to identify and characterise gentrification. Such data-driven approaches are possible because of the increased availability of data over small areas for fine spatial and temporal resolutions. They overcome limitations of traditional approaches to quantifying geodemographic change. This study uses annual data for 2010–2019 of House Price, Professional Occupation, Residential Mobility (in and out flows) and Ethnicity over small areas, Lower Super Output Areas (LSOAs). Areas of potential gentrification were identified from directional changes found in all of these variables, across combinations of start and end time periods. The initial set of areas were further processed and filtered to select robust gentrification cycles with minimum duration, and to determine start, peak and end years. Some 123 neighbourhoods in a regional case study area were found to have undergone some form of potential gentrification. These were examined further to characterise their spatial context and nature of the gentrification present, and specific types of gentrification were found to have specific periodicities. For example short-length durations (three to four years) were typically located in rural and suburban areas, associated with transit-induced cycles of gentrification, and greenification. Seven neighbourhoods were validated in detail, confirming the gentrification process and its type and their multivariate change vectors were examined. These showed that vector angle reflects the main data primitive driving the cycle of gentrification, which could aid with future prediction of gentrification cycles. A number of areas of further work are discussed.