ABSTRACT
San Francisco’s changing urban fabric has sparked meaningful debate between city officials, planners, architects, and residents over how to plan for the city’s continued growth. Skidmore, Owings, & Merrill’s research has contributed to this discourse through a set of three machine learning models that predict on a property-by-property basis the city’s susceptibility to change and future development. The research project used a curated, open-source dataset to train a machine learning model to predict the probability that any San Francisco property not in the development pipeline will appear in the development pipeline in the future. To intervene as designers, the team explored manually adjusting the weights of certain features and retraining the machine learning model to assess the impact of these changes on the prediction and likely trajectory of future development in San Francisco.
