An Intelligent Identification Method for Urban Renewal Potential Based on Spatial Gene Theory: A Case Study of Shanghai

Abstract

This study develops an intelligent method for identifying urban renewal potential using spatial gene theory and machine learning techniques. Addressing the limitations of traditional qualitative and static approaches to urban renewal assessment, the research proposes a four-step framework that identifies renewed spaces, extracts morphological evolution rules, constructs a spatial gene atlas, and predicts renewal potential. Using the China Building Rooftop Area (CBRA) dataset covering Shanghai between 2016 and 2021, the study analyses six morphological indicators derived from building footprints and applies Gaussian Mixture Models and Random Forest algorithms to identify spatial gene types and evaluate renewal potential. Seven spatial gene types were identified, revealing distinct morphological patterns across Shanghai’s urban fabric. The predictive model achieved an accuracy of 88.7%, with total building area, mean perimeter, construction year, distance to the city centre, and housing price emerging as the most influential variables. Results indicate that renewal potential is concentrated in historic inner-city districts and selected suburban centres. The paper argues that spatial gene identification can serve as a governance tool supporting adaptive, evidence-based urban renewal strategies while also highlighting the need to balance algorithmic approaches with public participation and institutional safeguards to ensure inclusive urban governance.