Research On The Evaluation Model Of Customer Risk Based On PSO-BP Neural Network For Car Insurance

With the rapid development of motor vehicles,people have great changes in the way of travel and the dependence on vehicles is increasing.Among them,the auto insurance as a kind of protection of venture capital investment,which with 70%of the amount of business accounted for the first place of property insurance companies,is attracting the attention of the insurance company.However,in the increasingly fierce competition in the insurance company,the entire auto insurance business management system is still not perfect,that resulting in the insurance company to pay too much for the customer,and the whole industry profits decreased significantly.So,it effectively uses the data that growing might of car insurance customer for the level of risk assessment model of auto insurance customers.the evaluation model of customer risk for car insurance is not only identify the potential risk of the customer in time,to provide customers for advice;but also provide a basis for financial research,social management,which has great social and commercial value.To this end,this paper research work as follows:1.Overview to the evaluation theory of customer risk related.It analysis an overview of the risk control theory,emphasizes the importance of risk control,a detailed analysis of the three categories of car insurance customer risk influence factors:risk factors vehicles,drivers' risk factors and environmental factors,to quantify the risk weighted index data processing,set up the whole car insurance customer risk assessment index system.Moreover,lists the current statistical model of risk assessment method to the customer and the artificial intelligence method.2.Research of BP neural network and particle swarm optimization(PSO)algorithm.It detailed introduces the theory,the idea,advantages and disadvantages of BP neural network and PSO,and an improved PSO algorithm which in view of the weak local searching ability of the PSO algorithm and the premature convergence problem based on the development of PSO.Last,it verifies the validity of the improved algorithm through standard function.3.Construction of BP neural network based on improved PSO for grade evaluation model of car insurance customer risk.It views the weak of the BP neural network model parameter and analyzes the PSO optimization principle for BP neural network.Furthermore,it use the improved PSO algorithm to optimize the weights and threshold of neural network,and construct for grade evaluation model based on improved PSO-BP to car insurance customer risk.Based on the research above,it use customer risk assessment data of some car insurance industry for the instance simulation,and the experiment results compare with the traditional BP model and basic PSO-BP model.The compared result to show that the proposed model has a faster convergence speed and higher accuracy,and verified this model is feasibility and practicability in the field of customer risk assessment.