A Study on Insurance Rate Decision Making in Disaster Frequent Areas Based on Pearson Correlation Analysis and Random Forest Algorithm

In recent years, earthquakes, floods, droughts, extreme heat, and other natural disasters have occurred more and more frequently in all corners of the earth, and these disasters have caused huge economic losses. The pressure on insurance companies to pay out in natural disaster-prone areas is much greater than in other areas, to make insurance companies operate normally, they need to set insurance rates according to local conditions, to set reasonable insurance rates, this paper obtains the highest correlation between the total disaster loss and the insurance loss through the Pearson's correlation analysis, and then adopts the Random Forest Algorithm, to obtain the highest weight for the number of people affected by the corrected disaster, and then adopts the Maximum Minimum Distance algorithm to classify all models, introduce the disaster damage factor (DDF) and normalize it, and use the BP neural network regression model to establish the evaluation model of the relationship between premiums and the impact factor, and the data visualization results show that the model fit is intuitively good. Combined with the theory of the time value of money, the insurance actuarial model was established from the perspective of insurance companies, and the model was validated by inputting relevant data from Guizhou China, and Brazil.