The Parameter Estimation Of Mixed Poisson Distribution Model

With the wide development of the mixed Poisson distribution in the medical, financial and insurance applications, it is receiving increasing attention. But, according to the author to understand, at present, studies on the Mixed Poisson distribution of the literature are relatively small, the most important reason is that it does not have a Mixed Gaussian distribution of a wide range of applications. However, with advances in computer technology and the development of statistical techniques, Mixed Poisson distribution analysis of statistical data will play an increasingly important role, thus the system in detail study of Mixed Poisson distribution model parameter estimation is necessary.The main achievements contained in this dissertation are as follows:1.We use moment estimation method, clustering method, EM algorithm to estimate parameters of Mixed Possion distribution in different conditions. Besides, we use Louis algorithm to estimate confidence interval.2.The simulation study shows that the more the differences between the means of the components, the less the differences between the proportion from Mixed Poisson distribution, the better the sampling and the fitting are, the estimated value is close to the theoretical value; Otherwise the effect is worse.