Research On Accumulated Claims Of Auto Insurance Based On Neural Network Algorithm

With the continuous advancement of my country's auto insurance rate reform,the position of insurance companies in auto insurance pricing has become more and more important.A reasonable forecast of premiums is very important to the company's steady development.The key to determining the rate is to calculate the pure premium.The core of calculating the pure premium is to calculate the cumulative claim amount.Therefore,the research on the cumulative claim amount is the most important thing.weight.At present,the mainstream method used in our country is the generalized linear algorithm,but it has its own limitations such as making assumptions about the data distribution in advance and being independent of each other.It will appear to be incapable of dealing with increasingly complex data,resulting in accumulated claims.The prediction results will have large deviations.Therefore,this article will model the cumulative auto insurance claim amount based on neural network and compare it with the generalized linear model.The main research contents are as follows:This article first introduces the accumulative claims made by domestic and foreign scholars in the generalized linear model and neural network model,and introduces the shortcomings of the generalized linear model itself and the necessity of using the neural network model.Then the generalized linear model and neural network model are introduced and theoretical models are constructed,and empirical analysis is carried out through experimental data to predict the cumulative claim amount,and the model is analyzed according to the selected standard.The results show that the neural network model is in The prediction accuracy of the cumulative claim amount has been significantly improved.Although the neural network model requires manual adjustment of parameters,and the calculation process is a black box without good interpretation ability,it improves the accuracy of prediction and can deal with more complex data environments,so it is beneficial for insurance companies to improve their own economic benefits.It also provides a certain reference value for the determination of auto insurance rates.