Probability-statistical properties of a two-parameter family of Poisson distributions of order \(k\) (Q1348540): Difference between revisions

The Poisson distribution of \(k\)-order with a generating function of the form \[ P_{\zeta}(z) = E\{z^{\zeta}\} = \sum_{n=0}^{\infty}z^nP\{\zeta=n\}= \exp\biggl\{-k\lambda + \lambda \sum_{j=1}^{k}z^j\biggr\}, \tag{1} \] where \(|z |\leq 1, \lambda > 0, k = 1, 2, \dots\;\), is considered. The distribution can be interpreted as a solution of the probability problem: let \(\zeta_i\) be \(k\) independent random variables which have the Poisson distribution with a parameter \(\lambda\): \(\zeta_j \approx P_0(\lambda)\); the problem is to find a distribution of the random variable \[ \zeta = \zeta_1 + 2\zeta_2 + \dots + k\zeta_k, \quad \tag{2} \] where \(k\) is given so that \(P(z)\) depends on the parameter \(\lambda\) only. The problem (1), (2) is generalised for a case of two parameters. Probabilistic properties of the obtained two-parametric distribution are studied and the methods of point-like estimates of parameters are disscussed.