Why does CB Predictor never select seasonal methods for my seasonal data?

There are several reasons why CB Predictor doesn't always select seasonal methods as the best for your seasonal data. Assuming your data is truly seasonal, to use the seasonal methods, you must set a seasonality on the Data Attributes tab.

If you do not select the seasonality option on this tab, CB Predictor doesn't even try to use the seasonal methods. Setting the correct seasonality of the data is also crucial. Use the smallest season possible. If you think your monthly data exhibits seasonality every quarter and every year, use 3 months for your season.

Also, the seasonal methods require two full seasons of historical data. So even when you set a seasonality, if you have less than two seasons of historical data, CB Predictor turns off the seasonal methods.

Another problem that keeps CB Predictor from selecting seasonal methods for seasonal data is that nonseasonal methods might better fit your data, even if the forecast of a straight line clearly seems like an inferior solution. When this occurs, you can override the selected best method in the Preview dialog before generating any output. To prevent this, you can try turning off the nonseasonal methods in the Method Gallery before running the forecast on your seasonal data, but don't try this if you are forecasting seasonal and nonseasonal time series together.