Time-varying auto-regressive models for count time-series

Count-valued time series data are routinely collected in many application areas. We particularly motivated to study the count of daily new cases, arising from COVID-19 spread. First, we propose a Bayesian framework time-varying semiparametric AR(p) model for and then extend it more sophisticated INGARCH model. calculate posterior contraction rates proposed methods with respect average Hellinger metric. Our structures models amenable Hamiltonian Monte Carlo (HMC) sampling efficient computation. substantiate our by simulations that show superiority compared some existing closely fit this setting. Finally, analyze newly confirmed cases NYC spread COVID three months.