Periodic ARMA model: Forecasting, parsimony, asymptotic normality and AIC.

摘要

Periodic autoregressive moving average (PARMA) models are indicated for time series whose mean, variance, and covariance function vary with the season. In this thesis, I develop and implement forecasting procedures for PARMA models. The required computations are documented in detail. An application to monthly river flow forecasting is provided.