Abstract
ABSTRACT This paper introduces a class of mixed integer-valued autoregressive model with dynamic mixing probabilities, which deals well with complex nonlinear structures and high-order dependencies of count time series. Model properties including stationarity, ergodicity and some conditional moments are studied. The model parameters are estimated using conditional least squares (CLS) and conditional maximum likelihood (CML) methods. Forecasting problems, including point forecasts and interval forecasts, have also been taking into consideration. The estimation effect is verified via numerical simulations. Finally, the proposed model is applied to two real data sets of crime counts, considering the driving effects of endogenous and exogenous explanatory variables, respectively.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call