Monitoring forecast model performance

You use forecast performance metrics to determine which model produces the best fit for the forecasted data.

1. Open the forecast that you want to test.
2. Open a forecast in the Forecast editor.
3. Run the forecast model that you want to test.
   The forecast results are displayed in the Results editor.
4. Click the Create Performance toolbar button.
   A table containing the forecast model's performance metrics is displayed.

   Cumulative Forecast Error

   Equal to the sum of differences between predicted and actual values.

   Mean Absolute Deviation

   Equal to the sum of the absolute values of the forecast error divided by the number of values. This metric tends to provide the best indicator of performance and is used as the default comparison criterion in forecast graphs.

   Mean Square Error

   Calculated as the sum (or average) of the squared error values. This performance metric is very sensitive to unique or large values, hence the error is amplified.

   Mean Absolute Percent Error

   Calculated as a percentage of the absolute difference between predicted and actual values divided by the number of values.

   Tracking Signal

   Calculated as a ratio of cumulative forecast error to the mean absolute deviation.

   In general, the closer the error is to zero, the better the performance of the model (for example, a performance error equal to zero implies a perfect fit between the predicted and actual values).

5. Optional: If you want to view the forecast hierarchical structure, open the Outline view by selecting Window > Show view > Outline from the main menu.