From diagnosis to action: Unveiling provincial-level levers for building carbon reduction in China via spatio-temporal clustering and grey-relational screening

Facing the "one-size-fits-all" dilemma in China's building-sector carbon governance, we propose a replicable spatio-temporal targeting framework that translates diagnostic analytics into actionable mitigation levers. Using 2006–2022 data for 30 provinces, we first quantify carbon-reduction performance through an economy-adjusted Tapio index, then map inter-provincial spillover channels via a modified gravity-SNA network. Temporal clustering employs two-stage temporal clustering (DTW-Ward + shape K-means), while spatial clustering optimises K-Medoids centroids by Harris Hawks Optimisation on SNA centrality indicators. The intersection yields four archetypes: Potential-Core, Potential-Periphery, Accumulation-Periphery and Pressure-Periphery. Grey-relational screening extracts the top-three drivers for each archetype and feeds them into differentiated policy bundles that couple pressure gradients with network centrality. Core provinces (Jiangsu, Guangdong, Shandong) function as network stabilisers and technology exporters; peripheral provinces receive tailored fiscal, certification and freight instruments to offset lock-in risks.