On construction and simulation of count data models (Q2479442): Difference between revisions

From the author's abstract: ``This paper develops a method for constructing nonnegative integer-valued random variables with any interval domain, any theoretically possible mean-variance pair, and different shapes.\dots As an application, our method enables production of count data models with full under- and over-dispersion flexibility and desired shape.'' The work contains: a mean-preserving discretization procedure for random variables; so-called `perfect initial models', based on beta and gamma random variables; two variate generation algorithms of count variables, based on discretizations of perfect initial models; a brief comparison with \textit{A. J. Walker} `alias method' [ACM Trans. Math. Softw. 3, No.~3, 253--256 (1977)]; and two open problems.