Publication

Abstract : For analyzing the count data, traditional probability distributions such as Poisson and negative binomial distributions are considered to be the most suitable models. But in some situations, count data shows large number of zeros that cause heavy tail which leads to over dispersion. In these situations, it is observed that these traditional statistical count models cannot be used efficiently. In order to overcome this problem, many mixed distributions have been introduced in the statistical literature. Among these distributions, Poisson and negative binomial were used as base line distribution for analyzing over dispersed count data. In this paper, we have proposed a mixture of negative binomial mixture distribution with generalized Lindley distribution and the resulting distribution is named as negative binomial-generalized Lindley (NB-GL) distribution. Further we have obtained some vital characteristics of the distribution such as mean, variance and factorial moments. Also we used maximum likelihood estimation method for estimation of parameters of proposed distribution.