| Popis: |
Generalized Space Time Autoregressive (GSTAR) model has an assumption of constant variance (homoscedastic). In this paper we described the development of GSTAR model by a variance of error is not constant, it changes over time (heteroscedastic), hence the term is heteroscedastic GSTAR model. The variance is modeled as the ARCH or GARCH models. Parameter estimation of the model used maximum likelihood method. The least squared method used to estimate the mean equation parameters, then the error terms parameters estimated by maximum likelihood method. |
