Research on Extreme Weather Risk Prediction Based on LSTM Model and Random Forest Algorithm

This paper focuses on how insurance companies can use Random Forest and LSTM models to make underwriting decisions, model real estate location and identify protected buildings when faced with the challenges of severe weather and frequent natural disasters. Firstly, the underwriting decision function model is established using indicators such as credit loss index (CRI), loss ratio, shortfall ratio, number of houses and risk loss, the TOPICS model based on entropy weighting method is used to determine the weather risk index, the risk loss is predicted using the LSTM model, and the value of CRI is determined and calculated by the entropy weighted topology method. Secondly, the Random Forest algorithm is used to establish a payout model to predict the insurance company's payout in the region, which assists the insurance company to formulate underwriting strategies in the region to ensure the stable operation and sustainable profitability of the insurance company. These research results provide a reference basis for insurance companies to formulate underwriting strategies, property development planning and cultural heritage protection.