Prediction of the impact of agricultural insurance on production stability based on machine learning

Based on machine learning methods, this paper studies the impact of agricultural insurance on production stability and its mediating mechanism of price and risk. Using XGBoost and LSTM models, the direct effect of agricultural insurance participation on yield volatility is analyzed, and the mediating effect of price volatility and risk perception is verified by step-by-step test method. The results show that agricultural insurance significantly improves production stability, with an importance score of 0.35, and yield volatility gradually decreases after insurance participation; price volatility and risk perception contribute 16% and 20% of stability improvement respectively, with significant indirect effects. The study reveals the mechanism by which agricultural insurance indirectly smoothes yield fluctuations by stabilizing prices and reducing risk perception, providing theoretical support and practical basis for agricultural risk management.