Research On Auto Insurance Fraud Recognition Based On Neural Network

With the gradual increase of the number of private car ownership,the case of auto insurance fraud is also on the rise.China's insurance industry started late,so the insurance industry can not effectively identify more auto insurance fraud cases,resulting in a lot of fraud cases that are difficult to detect,which has brought great economic losses to the insurance industry.Therefore,it is an urgent problem to find an effective method and a discriminant model to identify auto insurance fraud.Based on this,this paper uses Probit model and BP neural network model to analyze auto insurance fraud.First,the data is collected and organized.The identification of China's auto insurance fraud research and empirical analysis based on the latest research results at home and abroad.Then,this paper makes an empirical analysis on the claim data of 311 insurance sample cases(including 207 honest cases and 104 fraud cases)of B Branch of A listed property insurance company A in China.According to the collected case sample data of specific types of auto insurance,the data are firstly classified,and the model factors are selected according to the previous research conclusions.The empirical analysis is carried out by using Probit regression analysis method to extract the effective identification factors with the significance of the model.The refined recognition factors and the unrefined one are respectively used as input data to construct the neural network model.It is found that the efficiency of refined factors is significantly higher than that of the unrefined factors.Using these significant factors as input data,the neural network model is trained and predicted,and its validity is tested.At the same time,in view of a series of problems such as the BP neural system network technology's high dependence on enterprise samples and slow convergence speed,this paper adopts the sparrow algorithm to optimize the neural network to solve this problem and overcome the difficulties.Finally,this paper put forward some countermeasures and targeted suggestions for auto insurance fraud.