老旧小区改造优先级评估机制与决策支持研究OA

In the context of China's urban transition from large-scale expansion to intensive regeneration,the establishment of a scientific priority assessment mechanism for the renovation of old residential communities has become a critical yet challenging task in urban governance.This study aimed to address the prevailing challenges in contemporary decision-making processes,including the complexity of quantifying public needs and the absence of a comprehensive city-wide perspective.To this end,a novel data-driven priority evaluation framework was proposed using an empirical case study of Hangzhou.Methodologically,this study introduces key innovations.First,a Graph Neural Network(GNN)model incorporating spatial semantics was developed in order to identify the construction era of residential communities,achieving a high-precision recognition rate of 97.6％and effectively overcoming the limitations of traditional image classification.Second,a Large Language Model(LLM)is employed to perform a fine-grained sentiment analysis of geotagged social media texts,and a geographically weighted approach is applied to spatially quantify public sentiment at the neighborhood scale.Third,moving beyond conventional facility-oriented evaluation,an integrated three-dimensional indicator system was constructed,covering problem urgency,implementation feasibility,and potential benefits.Priority assessment was conducted using an XGBoost model,and its robustness was strengthened through Monte Carlo simulation and full-gradient sensitivity analysis,ensuring reliable and stable ranking outcomes across all 2 188 old residential communities under study.The main findings are as follows:(1)a distinct scale-dependent spatial heterogeneity in renovation priorities.While high-priority communities are clustered in central urban districts,the average priority score is notably higher in certain peripheral counties—an"inversion"phenomenon that reflects the underlying tension between spatial equity and renewal efficiency,shaped by historical investment gaps,diminishing marginal returns,and regional economic disparities.(2)A dual-scale spatial clustering analysis conducted at both the community and administrative district levels identified four renovation archetypes and four corresponding policy zones,offering a refined understanding of priority differentiation and enabling a targeted governance approach characterized by categorized renovation and zonal coordination.(3)Building on these spatial patterns,this study has proposed a synergistic governance paradigm that integrates differentiated renovation strategies with macro-level resource allocation,thereby enhancing the precision and effectiveness of policy implementation.By establishing a"data-driven,model-supported,policy-adaptive"technical pathway,this research contributes to shifting urban renewal practices from experience-based judgment to evidence-based decision-making.The framework not only provides a scientific basis for optimizing renovation sequencing and resource distribution but also translates complex spatial diagnostics into operable policy tools,supporting more equitable,efficient,and sustainable urban regeneration in high-density cities.