This dissertation develops a modeling framework for univariate and multivariate zero-inflated time series of counts and applies the models in a clustering scheme to identify groups of count series with similar behavior. The basic modeling framework used is observation-driven Poisson regression...