Extreme weather events in today's world are causing significant damage to people's property and affecting the decisions of insurance companies. We develop and optimize Insurance Risk Assessment Model (IRAM) and Community Building Preservation Model (CBPM) models to quantify the risk of increasing insurance policies and provide recommendations for insurers and real estate companies to make decisions. We collect data from two countries, then select indicators and use the AHP-EWM approaches and Topsis method to derive the weights of each indicator and calculate the composite score, then we use a gray prediction model to predict the Insurance risk comprehensive score (IRC) of the future direction and set up a benchmark to predict the suitability of increasing the number of contracted policies. To solve the problem of whether and how to establish a real estate project, we construct an objective function and formulate restrictive conditions, and use Particle Swarm Optimization to find the optimal approach to establish the real estate project. Moreover, we select specific historic building indicators and combine them with IRC to train a machine classification model by using the XGBoost method, which can label specific neighborhood buildings as deserving of additional policy protection. Finally, we select Himeji Castle as historic building and evaluate it using IRAM and CBPM respectively, the result of IRAM shows that the region where Himeji Castle is located is not suitable for increasing contracted policies, and the result of CBPM shows that Himeji Castle is unsuitable for continuing to add policies to be protected. Our model IRAM and CBPM pass the sensitivity analysis, and final prediction results show that the accuracy rate and recall rate of the classification model are 0.91 and 0.88, which has the best performance in terms of each metric used for comparison.
|