Model Selection in ARMA(p,q) Processes
In practice, the form of the underlying process that generated the data is unknown. Should we estimate an AR(p) model, an MA(q) model, or an ARMA(p,q) model? Moreover, what lag lengths of p and q should we choose? We simply do not have good a priori reason to suspect that the data generating process is of one type or another, or a combination of the two. How is a researcher to proceed? Which sort of model should we estimate?
- Box, G. E., & Jenkins, G. M. (1976). Time series analysis: Forecasting and control (revised ed.). Oakland: Holden-Day.Google Scholar