|
Abstract:
|
A popular special case of the finite mixture models is a finite mixtures for the count data. For the case of multiple points of inflation in the small counts, where a clear theoretical relationship between small counts may exist, people have employed the finite mixtures of truncated Poisson distributions. For such models, the identifiability is an issue that must be addressed before any inference about parameters. In this work, based on Teicher's results on the identifiability of finite mixtures, by using the probability generating function, the conditions on identifiability for a finite mixtures of truncated Poisson distributions are studied.
|
