Font Size: a A A

The Research On The Loss Of Insurance Based On The Bayesian Mixed Exponential Distribution

Posted on:2017-04-01Degree:MasterType:Thesis
Country:ChinaCandidate:Y ZhangFull Text:PDF
GTID:2359330515481384Subject:Applied Statistics
Abstract/Summary:PDF Full Text Request
Loss is one of the most important problems in the insurance company actuary measures and non life insurance actuarial research important topic,on behalf of the risks faced by insurance company,the correct identification of the insurance companies,the risk of loss distribution of insurance companies is of great significance,for the daily management of the company,the evaluation of assets and liabilities,cash flow forecast,product development,and has a decisive role,select the appropriate loss distribution of insurance companies is very important.The development of China’s property insurance is still in its infancy,non life insurance business especially of new business,we cannot through business came to a large number of data,are only in the last few years of observation samples to estimate the loss distribution no longer has excellent characteristics,under the condition of small sample traditional classical statistical estimation may lead to large deviations,Bayesian method overcomes the shortcomings of insufficient historical data,using prior information to get more accurate estimates,has been widely used in the field of insurance actuarial.Based on the Bayesian mixed exponential distribution model as the foundation,building loss assessment model.Occurred from the loss amount and loss probability of two point of view,the use of Bayesian estimation method,using Markov chain Monte Carlo simulation(MCMC)of Gibbs sampling,through data loss distribution fitting parameters is estimated.In the parameter estimation of the amount of the loss,Gibbs sampling than other Bayesian approximation estimation algorithm is more robust;in loss probability estimates,at a given default distribution based on,MCMC algorithm solves the complex Bayesian computational problems,the existing observational data,the weight is constantly updated,more hasten is closer to the real weight loss.For the accurate estimation of the two,can provide the basis for the management decision of insurance company.
Keywords/Search Tags:Loss, Mixed Exponential Distribution, Bayesian Estimate, MCMC Algorithm
PDF Full Text Request
Related items