| Summary: | Gentrification is a contentious issue which local governments struggle to deal with because warning signs are not always visible. Unlike current literature that utilises solely socio-economic data, we introduce the use of large-scale spatio-temporal mobility data to predict which neighbourhoods of a city will gentrify. More specifically, from mobility data, which is associated with the exchange of ideas and capital between neighbourhoods, we construct mobility networks. Features are extracted from these mobility networks and used in gentrification prediction, which is framed as a binary classification. As a case study, we use the Taxi & Limousine Commission Trip Record Data to predict which census tracts would gentrify in New York City from 2010 to 2018, and show that considering network features alongside socio-economic features leads to a significant improvement in prediction performance.
|
|---|