Can Generalized Poisson model replace any other count data models? An evaluation

BackgroundCount data represents the number of occurrences an event within a fixed period time. In count modelling, overdispersion is inevitable. Sometimes, this may not be just due to excess zeros but presence two or more mixtures. Hence main objective examine for mixtures if any, with and compare Generalized Poisson model, Mixture models other using real time simulated data.MethodsThree over-dispersed datasets were used comparison models. The compared information criteria like AIC BIC regression coefficients. Data was also mixture zeros. simulation repeated different sample sizes identify better model.ResultsGeneralized showed consistently lower bias MSE when varying sizes. values almost similar Poisson, ZIP model. Similar findings obtained from data.ConclusionGeneralized provides fit overdispersed zeros, in Negative Binomial can redistricted re-evaluated against