Modelos para dados de contagem: um estudo sobre o número de ovos do mosquito Aedes aegypti

Abstract

In this work we considered four types of models for count data: the Poisson model, the Negative Binomial Model and variations of these models for count inflated of zeros (ZIP and ZINB). Information on the number of put eggs for the mosquito aedes in the city of São José of Rio Preto, SP, was analyzed through models for counting data. The number of eggs of the mosquito aedes is studied firstly without the covariable and through some tests, it was verified that models zero inflated (ZIP and ZINB) are more adapted to analyze the data under study. The ZIP model was shown inadequate, due to overdispersion generated by the variability of the data. The ZINB model, was shown more flexible to the data as it is shown graphically.