Analysis of Univariate Forecast Errors
In this process, you improve the accuracy of your univariate forecasts by monitoring predefined tolerance thresholds for the standard errors, and by adjusting the forecast model where any of these thresholds are exceeded.
You have installed the Alert Monitor.
In the univariate forecast profile, set which
forecast error measurements you want to be calculated.
In the diagnostic group which you enter in the univariate profile, specify the upper limits of the forecast error measurements.
In the Alert Monitor, create a forecast alert profile.
Alert profiles are used to display specific alerts to specific users. They do not affect whether or not alerts are created.In interactive demand planning (for mass processing, see * below), assign the alert profile you created in step 3 by choosing Edit → Assign alert profile. If desired, select Forecast deletes old alerts automatically.
Run the forecast in the statistical forecast view.
The forecast errors that you set in the univariate profile are calculated automatically.
You can view them on the Forecast errors tabstrip of the statistical forecast view. To compare the forecast errors with those of other models, look at the different forecast versions by choosing Settings ® Forecast comparison. From the workspace toolbar, choose Alerts on/off .
Check the alerts to see if any forecast errors have exceeded their predefined upper limits.
Make any necessary adjustments to the univariate forecast model.
Run the forecast again.
Study the alerts to see if the adjustments you made in step 8 have corrected the problem.
If necessary, repeat steps 8 through 10.
* If you are forecasting in the background, continue as follows at step 4.
- Specify in the mass processing activity that alerts are to be created.
- Run the forecast in the background as a mass processing job.
- Open the Alert Monitor by choosing Supply Chain Monitoring ®
Alert Monitor from the SAP Easy Access menu.
- Check the alerts.
- If necessary, make adjustments to the univariate forecast profile or use a different profile.
- Repeat steps 5 through 8 as often as necessary.