Research on spatiotemporal evolution patterns and driving mechanisms of material cultural heritage based on machine learning

Tangible cultural heritage serves as a vital testament to the evolution of human civilization, bearing profound historical memories. This study focuses on China’s Shu Road, integrating multi-source historical data and employing GIS and machine learning methods to reveal the spatiotemporal evolution and driving mechanisms of material cultural heritage along its route. Results indicate that from the pre-Qin period to the Ming and Qing dynasties, heritage distribution gradually shifted southward from Shaanxi to Sichuan. The Ming and Qing periods marked a peak in both quantity and concentration, with significant growth observed in settlements and religious heritage. Different heritage types exhibit spatial variations. Taking settlement heritage as an example, the CatBoost-SHAP model identified population density, distance to rivers, and other factors as key, with all factors showing nonlinear effects and complex interactions. This study deepens our understanding of the evolution patterns of Shu Road cultural heritage and provides scientific basis for its conservation.