كاربرد نظريه باورمندي در پيشبيني تعداد خسارتها در بيمه اتومبيل

6. Application of the Credibility Theory to Claim Frequencies Prediction in Automobile Insurance

The current paper breaks down into two main parts. The first part develops a Markov Chain Monte Carlo (MCMC) algorithm in order to test the statistical hypotheses regarding three discrete distributions commonly used to model the claim counts in actuarial science. The proposed methodology involves advanced statistical techniques of Bayesian modeling which makes use of the Gibbs sampling variable selection algorithm. The most important advantage of this approach is its straightforward implementation using the Win BUGS software. The second section involves credibility theory for claim frequencies prediction in future periods. Finally, in the last section, both methodologies are applied to the real claims data set of automobile insurance contracts of an Iranian private insurance company (Parsian Insurance co.) and the claim frequency predicted in the year 1386.