A new first order mixed integer valued autoregressive processes for time series with Poisson innovations
INAR(1), INARCH(1), binomial thinning, Poisson thinning, time series.
Time series, viewed as a collection of observations measured sequentially over time, has been studied with deep notoriety in recent years, based on applications and new proposals of autoregressive models that extend the field of study. This paper proposes a new first order mixed autoregressive model with Poisson innovations, denoted POMINAR (1), mixing two operators known as thinning binomial and thinning Poisson, it also provides the interpretation of these operators and demonstrates their respective properties, as well as a possible case of applying POMINAR (1). Expected value, variance, conditional expectation and conditional variance of the proposed process are taken step by step. In detail the transition probabilities are developed. The maximum conditional likelihood estimators to the process parameters are determined and an application to a real data set is given seeking the effectiveness of the proposed model.