Example of Time Series Forecast
Use the Time Series Forecast platform to fit a variety of state space smoothing models to data with multiple time series. The final model is selected through the Forecasting Performance option, which utilizes a holdback set.
1. Select Help > Sample Data Folder and open Time Series/M3C Quarterly.jmp.
2. Select Analyze > Specialized Modeling > Time Series Forecast.
3. Select Y and click Y.
4. Select Series and click Grouping.
5. Select Time and click Time.
6. Click OK.
The initial Time Series Forecast report appears. It contains summaries of the Time and Y variables. There is also a Modeling Specifications report that provides options for fitting forecast models.
Figure 19.2 Initial Report for Time Series Forecast
7. In the Model Specifications report, click the Complete Specifications tab.
8. Click the menu below Model Selection Strategy and select Forecasting Performance.
9. Click Run.
This fits a set of recommended models to the data. The best model for each level of the grouping variable is chosen and reported in the Model Reports. See Model Reports.
Figure 19.3 Model Reports
The model report for one series, Series N646, is shown. This corresponds to the series name that is selected in the Select Series list. You can use the down arrow key to scan through the model reports for each series. Alternatively, you can click a different name in the list to display the model report for the selected series.