Study on insurance in earthquake-prone areas based on BP neural network and logistic regression model

Abstract

The frequent occurrence of extreme events poses major challenges to the insurance industry and affects its sustainable development. To provide insurance recommendations for extreme disaster areas, this paper combines BP neural network model, principal component analysis method and Logistic regression model to establish our insurance model. First, we processed the data. Then, the BP neural network mode used to correlate 33 kinds of earthquake correlation indexes with earthquake probability and make prediction. To make the data more representative, we conducted dimension reduction prediction through principal component analysis, screened out 8 important factors, and identified 3 principal components. We then performed Logistic regression analysis based on these three principal components and used the Logistic model to decide whether to conduct insurance. After the experiment, the accuracy of the model can reach 78.9%, which can provide a reliable prediction of the disaster in the region, to help the insurance company to make a decision whether to insure in the region.