Robert Lund, University of Georgia, Athens


Periodic Time Series


Many time series display periodicities in their mean and/or
autocovariances.  This talk overviews general properties of periodic
series and discusses some statistical issues in the analysis of periodic
time series data. Comparisons of autoregressive moving-average (ARMA) 
series, seasonal autoregressive moving-average (SARMA) series, and
periodic autoregressive moving-average (PARMA) series will be made. 
Likelihood computation, recursive prediction, large sample theory, and
parsimonious model fitting for PARMA series will be explored.