Characterising neighbourhood dynamics through social media anlaysis and house sales transactions
Keywords: neighbourhood dynamics, house price model, Twitter
Abstract. This paper describes a two stage approach for identifying neighbourhood areas that may be undergoing gentrification related changes. It summarises classic hedonic house price data over time (2014–2023) for each neighbourhood, and compares neighbourhood average price with those of local nearby areas. This enables neighbourhoods experiencing high relative increases in price to be identified as potentially gentrifying areas. Social media data for these areas were extracted and analysed using a large language model which scored each individual social media post by the degree to which their content indicated that the neighbourhood is experiencing change, potentially providing confirmatory evidence or not of gentrification. A number of areas of further work are identified.