Recording – Exploring Multi-Dimensional Nature of Gentrification Using Machine Learning

Join us for a comprehensive webinar based on the research paper from the 2023 Peter Barrington APREF Award winner, William Thackway.

William’s thesis investigates the use of machine learning methods to predict and validate gentrification hotspots in Sydney.

The thesis is comprised of two research papers documenting different machine learning models: the first a tree-based model that predicts future gentrification in Sydney up to five years forward using socioeconomic and housing data, and the second a deep learning model that validations gentrification predictions by identifying residential property upgrades using Google Street View images. The thesis provides the first machine learning approach to combine both social and physical indicators of gentrification. Ultimately, the predictive tool developed within the thesis can help support social policy interventions to better mitigate the impacts of gentrification-induced displacement.

Gentrification is a multi-dimensional urban change process which involves both socioeconomic and physical changes to a neighbourhood. Machine learning enables analysis of new and complex data sources compared to traditional modelling techniques. This can enrich our understanding of urban change processes. Investigating multiple components of an urban change process can provide a more comprehensive and validated understanding of that process.