After we have looked at a possible concept for a rolling forecast in SAP Analytics Cloud in the blog post "SAP Analytics Cloud Update of Actuals in Rolling Forecast", I will first insert a projection into our forecast scenario in an intermediate step. SAP Analytics Cloud provides two options for this:
- Predictive Forecast (Automatic)
- Predictive Scenario (with training options on data).
Here we will use a Predictive Scenario (Time Series Predictive Scenario) to populate the projection and version PF1.
Figure 1: Concept for rolling forecast with ACTUAL (Actual), RF1 (Forecast 1) and PF (Projection). Before the forecast (RF1) is entered manually, we "post" the automatic projection (PF).
In Figure 1 we see our ACT and the two scenarios (SAP Analytics Cloud "versions") RF1 and PF. PF stands here for the "Predictive Scenario". Figure 2 shows the menu entry in SAP Analytics Cloud for this function.
From month to month the ACTUAL is updated and the "Cut Over Date" is set one month further. To do this, we optionally create a new forecast version for each month (for example: RF1, RF2, ...RF[n]).
Before we hand in our RF1, we want to offer a suggestion value to our participants in the rolling forecast process. This not only provides a convenience to the planners, but also allows for a shortening of the forecasting process. Accurate extrapolation can save considerable effort in manual planning. Especially in the case of a large number of products, articles or customers, for example, automatic extrapolation offers great potential for reducing the forecasting effort and speeding up cycles.
Figure 1: Menu "Predictive Scenario" in which the time series projection is created with MR
Set up projection by an employee, e.g. data scientist using SAP Analytics Cloud Predictive Scenario.
Figure 3: Projection ("Predictive Scenario") based on planning model and data in SAP Analytics Cloud
Before the transfer of the extrapolation results, the scenario or version "PF1" and "RF1" is empty. In Figure 4 we see the rows, but from the cut-over date January 2021 only empty. IST (ACT) data is available up to December 2020. We loaded the data in the last blog post and set the cut-over date.
Figure 4 Actual (ACT), Rolling Forecast (RF1) and Projection (PF1). RF1 and PF1 are not yet initiated (still empty).
Figure 5 Training on the data of the planning / forecasting model in SAP Analytics Cloud
Figure 6 After the projection has been generated, it can be transferred to the Forecasting Scenario (in the example the version or scenario PF1)
Figure 7 The extrapolation has been transferred. Now, the extrapolation can be also copied to an RF version and then serves as a starting point for further processing.
Figure 8 Transfer of projection to RF1 (rolling forecast)
Now the next forecasting cycle can begin and the participants in the process enter their manual adjustments in the rolling forecast only where necessary. In the best case, these are only a few occurrences, e.g. for strategic products or exceptional situations.
- Predictive Forecast (Automatic)
- Predictive Scenario (with training options on data).
Here we will use a Predictive Scenario (Time Series Predictive Scenario) to populate the projection and version PF1.
Figure 1: Concept for rolling forecast with ACTUAL (Actual), RF1 (Forecast 1) and PF (Projection). Before the forecast (RF1) is entered manually, we "post" the automatic projection (PF).
In Figure 1 we see our ACT and the two scenarios (SAP Analytics Cloud "versions") RF1 and PF. PF stands here for the "Predictive Scenario". Figure 2 shows the menu entry in SAP Analytics Cloud for this function.
From month to month the ACTUAL is updated and the "Cut Over Date" is set one month further. To do this, we optionally create a new forecast version for each month (for example: RF1, RF2, ...RF[n]).
Before we hand in our RF1, we want to offer a suggestion value to our participants in the rolling forecast process. This not only provides a convenience to the planners, but also allows for a shortening of the forecasting process. Accurate extrapolation can save considerable effort in manual planning. Especially in the case of a large number of products, articles or customers, for example, automatic extrapolation offers great potential for reducing the forecasting effort and speeding up cycles.
Figure 1: Menu "Predictive Scenario" in which the time series projection is created with MR
Set up projection by an employee, e.g. data scientist using SAP Analytics Cloud Predictive Scenario.
Figure 3: Projection ("Predictive Scenario") based on planning model and data in SAP Analytics Cloud
Before the transfer of the extrapolation results, the scenario or version "PF1" and "RF1" is empty. In Figure 4 we see the rows, but from the cut-over date January 2021 only empty. IST (ACT) data is available up to December 2020. We loaded the data in the last blog post and set the cut-over date.
Figure 4 Actual (ACT), Rolling Forecast (RF1) and Projection (PF1). RF1 and PF1 are not yet initiated (still empty).
Figure 5 Training on the data of the planning / forecasting model in SAP Analytics Cloud
Figure 6 After the projection has been generated, it can be transferred to the Forecasting Scenario (in the example the version or scenario PF1)
Figure 7 The extrapolation has been transferred. Now, the extrapolation can be also copied to an RF version and then serves as a starting point for further processing.
Figure 8 Transfer of projection to RF1 (rolling forecast)
Now the next forecasting cycle can begin and the participants in the process enter their manual adjustments in the rolling forecast only where necessary. In the best case, these are only a few occurrences, e.g. for strategic products or exceptional situations.
- SAP Managed Tags:
