Discovering income-economic segregation patterns: A residential-mobility embedding approach

Abstract

As most studies of segregation rely on the evenness dimension, this current study proposes a graph embedding approach to explore the usefulness of employing the isolation-exposure dimension to evaluate income segregation. While most segregation studies analyzed the static distribution of population subgroups, current study attempts to classify neighborhoods based on house value as a proxy of income, residents' exposure to people of different income levels as constrained by their mobility patterns, and amenities available in the neighborhood. This study exploits the graph embedding method to classify neighborhoods by combining their various attributes, static population distribution and mobility data provided by smart cards to analyze income segregation in Shenzhen, China. Results identify four types of communities with different economic statuses, mobility patterns, and amenity characteristics. They provide rich descriptions about the connections between income segregation patterns, population dynamics, and neighborhood characteristics. The study found that the more segregated communities, which are composed of the poorest and richest groups, are mostly in the peripheral regions of the city while the inner city has lower levels of segregation, mainly due to differentials in transit accessibility. The study demonstrates the great potential of the proposed method to incorporate multiple aspects to evaluate segregation.