Selecting an ARIMA Model Selection Criterion

  To select an ARIMA model selection criterion:

1. In the Predictor wizard Methods panel, select ARIMA.

2. In the Autoregressive Integrated Moving Average (ARIMA) Details panel, select Automatic (the default).

3. Select Minimize Information Criterion, and then click Select Information Criterion (Alt+e).

4. In the Select Information Criterion dialog, select a setting:

   • Bayesian Information Criterion (BIC)

   • Akaike's Information Criterion (AIC)

   • Corrected AIC (AICc)

   Note:

   See the Bibliography for references that discuss the differences among these criteria. The three criteria differ in the way that they penalize overfitting. The differences are small, and the chosen criterion typically does not lead to a change in the ARIMA model selected as the best fit.