A New Flexible Discrete Distribution: Theory and Empirical Evidence
Giovanni Pollio, Giovanni De Luca

Abstract
A new discrete distribution with two parameters is introduced and discussed. We present its derivation after identifying the generator mechanism of the mass probability function. Then we show its flexibility in terms of dispersion index and show how to estimate the parameters by maximum likelihood. Finally, we compare it to traditional as well as flexible discrete distributions using some popular insurance datasets. The AIC and BIC criteria strongly suggest that the new distributions is able to provide a fit to discrete data in a very satisfactory way.

Full Text: PDF