Spatiotemporal dynamics and key drivers of resource recycling industry in China (1987–2024): A multisource big data approach and machine learning analysis

Understanding the spatiotemporal dynamics and key drivers of recycling enterprises is essential for optimizing resource recovery systems and advancing sustainable development in China. This study adopts a multisource big data approach, integrating geospatial, economic, and environmental datasets from 300 cities between 1987 and 2024, and applies spatial analysis and Random Forest models to examine 5,171 registered recycling enterprises. Results reveal strong spatial concentration in eastern coastal provinces like Jiangsu (over 500 enterprises), Shandong, and Zhejiang. Despite gradual westward expansion since 2010, the western region accounts for only 18.4% of the total. The industrial chain presents spatial heterogeneity: upstream and downstream enterprises are dispersed, while 65.5% of midstream enterprises cluster in the Yangtze River Delta. Random Forest analysis shows that patent grants (0.327) and retail sales (0.273) are the top national predictors. Regionally, innovation dominates in eastern (0.311) and central (0.500) China, with carbon emissions also influential (0.124 and 0.149), whereas market size leads in the west (0.497). Temporally, enterprise growth evolved from market- and labor-oriented drivers (2002–2009), to innovation-driven expansion (2010–2020), and finally to an environmental governance phase after 2021, where carbon emissions (0.219) became the primary spatial determinant. By leveraging big data and machine learning, this study provides insights for optimizing recycling networks and enhancing regional sustainability.