A study of property insurance based on ARFLGB-XGBoost modeling

Abstract

Extreme weather events have become a crisis for property owners and insurance companies, and insurance companies have changed the way they are willing to underwrite policies. The purpose of this report is to develop a comprehensive assessment model for the multiple factors that influence underwriting policies, in the hopes of providing the community with strategies for preserving historic landmarks in the future. In this paper, we first collected data on six climate hazard impact indicators in China and the United States after 2000, then preprocessed the nulls and outliers in the original dataset with linear and exponential fitting, and then evaluated the indicators using the CVM-CRITIC model. In order to analyze how community and demographic factors affect the application of the model, this paper uses the Spearman algorithm to analyze the correlation between community and demographic indicators and climate risk indicators, and uses the ARFLGB-XGBoost model for regression prediction in order to demonstrate the feasibility of the HICD model from different perspectives.