A Class of Copula-Based Bivariate Poisson Time Series Models with Applications

A class of bivariate integer-valued time series models was constructed via copula theory. Each follows a Markov chain with the serial dependence captured using copula-based transition probabilities from Poisson and zero-inflated (ZIP) margins. The theory also used again to capture between two either Gaussian or “t-copula” functions. Such method provides flexible structure that allows for positive negative correlation, as well. In addition, use permits applying different margins complicated such ZIP distribution. Likelihood-based inference estimate models’ parameters integrals t-copula functions being evaluated standard randomized Monte Carlo methods. To evaluate proposed models, comprehensive simulated study conducted. Then, sets real-life examples were analyzed assuming marginals, respectively. results showed superiority models.