شماره ركورد كنفرانس :
3140
عنوان مقاله :
Modeling overdispersed discrete time series
عنوان به زبان ديگر :
Modeling overdispersed discrete time series
پديدآورندگان :
Aghababaei Jazi Mansour نويسنده Faculty of Mathematics - University of Sistan and Baluchestan - Zahedan , Jones Geoff نويسنده Institute of Fundamental Sciences - Massey University - Palmerston North - New Zealand , Chin-Diew Lai نويسنده Institute of Fundamental Sciences - Massey University - Palmerston North - New Zealand
كليدواژه :
Binomial thinning operator , Conditional maximum likelihood estimation , Zero inflated Poisson distribution , Discrete time series
عنوان كنفرانس :
يازدهمين كنفرانس آمار ايران
چكيده لاتين :
The classie: first-order normnegative integer valueed autoregressive (INAR(1)) process is usually assumed to have Poisson marginal distribution and consequently, the corresponding innovations are Poisson random variables. This model is naturally suitable for equidispersed time series wherein the mean and variance are the same. In this paper, we consider modeling overdispersed discrete time series. At first, we propose a simple check for the distribution of innovations in INAR(1). Then, we fit INAR(1) model with each Poisson, geometric and Zero inflated Poisson innovations to two real overdispersed time series to illustrate the Superiority of INAR(1) model with geometric and Zero inflated Poisson innovations. We use conditional maximum likelihood approach for the estimation of the parameters and the predicted values of the time series.
شماره مدرك كنفرانس :
4219389
