An Individual Claim Reserving Evaluation Model Based On Neural Network Algorithm

Insurance is a product that the insurance company can transfer the possible loss of the enterprise or individual in the future by charging a certain amount of premium and realize the function of risk management.It is a powerful means to deal with risks in the market economy.This risk transfer is crucial in today’s society.As one of the important types of insurance,nonlife insurance is also the main business of insurance companies in recent years.With the growing maturity of "Internet +" technology,China has officially opened a new era of "artificial intelligence",and machine learning methods have brought huge impact and influence to various disciplines.At present,the research focus in actuarial field is to combine machine learning with traditional theoretical knowledge.Based on this,this paper aims at the evaluation of outstanding claims reserve in non-life insurance actuarial science,tries to build an evaluation model of individual claims reserve based on neural network,and explores the feasibility and practicability of machine learning method in non-life insurance reserve estimation.Focusing on the issue of individual claim reserve evaluation,this paper summarizes the overall research methods and content of this paper by combining the key issues of actuarial science research with the shortcomings of current research in this field.In the aspect of actuarial knowledge,this paper introduces the traditional claim reserve evaluation method based on aggregate data commonly used in practice.In the aspect of neural network algorithm knowledge,the background of the development of neural network technology and the concepts and methods used in this paper are briefly introduced.The method introduced in this paper realizes the statistical model of predicting the amount of individual outstanding claims reserve by using the policy data of individual paid claims and reported claims,and makes an empirical analysis with a group of real data of a French company.The analysis of experimental results shows that the individual claim reserve evaluation model based on neural network proposed in this paper can meet the adequacy requirements of reserve withdrawal,but the accuracy of the training model needs to be further improved for the evaluation of the year with less known data.