Optimization and driving mechanisms of agricultural resilience measurement based on XGBoost-SHAP model: Evidence from China

As agricultural systems face increasingly intertwined and recurrent external risks, strengthening the resilience of China’s agricultural industry has become a strategic priority for ensuring national food security and agricultural sustainability. A robust and systematic measurement of the resilience of China’s agricultural industry (RCAI), together with an examination of its principal drivers, is therefore essential. This study develops a comprehensive RCAI evaluation framework for 31 provincial-level regions in China, characterizes the spatiotemporal evolution of RCAI from 2000 to 2022, and employs an XGBoost–SHAP analytical framework to refine the index and uncover underlying driving mechanisms. The results show that: (1) the resilience of China’s agricultural industry exhibits a clear upward trend over time and a pronounced spatial pattern of “higher in the east and lower in the west”; (2) the XGBoost model optimizes the RCAI evaluation framework, yielding an index that is more robust and representative; (3) agricultural fixed-asset investment contributes 30.6% to RCAI, making it the most influential determinant of resilience enhancement. In addition, rural consumption expenditure and transportation infrastructure are positively associated with RCAI and display threshold effects. Overall, the findings demonstrate that the XGBoost–SHAP framework can effectively capture complex nonlinear relationships between RCAI and its determinants and can improve the precision of resilience measurement.