The forecast results depend greatly on which forecast model you use and how you set it up. The simulation and storage of multiple demand planning scenarios allows you to evaluate different forecast models and settings in Demand Planning before deciding what data to release to Supply Network Planning (SNP). Moreover, the simulation process in Demand Planning can be done in conjunction with the simulation of planning scenarios to drive S&OP. For example, you can release two demand planning scenarios to SNP for simulation purposes: One based on the assumption that there will be a sales increase of 5% in the next six months, the second based on the assumption that there will be a sales increase of 10% in the next six months.
A demand plan is stored in a planning version within a planning area. Within one planning version, there can be multiple forecast versions. For more information about forecast versions, see Forecast Comparison.
You have decided which forecast models you wish to use for which products. For more information, see Definition/Redefinition of Forecast Models.
Forecast simulation can be done at several stages in the demand planning cycle; that is, both to evaluate the forecasts of individual departments and to compare different consensus-based forecasting scenarios. Participants in the simulation process are therefore planners from sales, marketing, and logistics as well as supply chain managers.
The forecast simulation process consists of the following steps:
1. Create several forecast scenarios based on different techniques, models and parameters. For more information, see Creation of the Demand Forecast
2. Compare the different scenarios (check individual forecasts / check demand plan proposals).
3. Evaluate the quality of each version of the forecast through macros, the univariate forecast errors and the MLR measures of fit.
4. Release the best fitting scenario(s) to SNP. For more information, see Release of the Demand Plan to SNP.
5. Retain in Demand Planning the version(s) that you released as well as the most important versions that you did not release. Delete the less important versions; this helps enhance performance.
6. Check the accuracy of the released version(s) as well as on the retained, non-released versions. For more information, see Monitoring of Forecast Accuracy.
This process enables you to optimize the demand plan in each successive forecasting cycle.