We need to create a good and robust statistical forecast that we use for automatically creating our forecast projections. SAP IBP for demand offers a variety of different statistical forecast algorithms. We create one profile with different forecast algorithms and let the system pick the best algorithm.
Task 1: Create a Forecast Profile and Add Three Forecasting Algorithms
Steps
Go to the Manage Forecast Models app. Create a new forecast profile called PickBestModel## that picks the best of three different algorithms using the information from the following table.
Field Value Model Name PickBestModel## Description Pick Best Model ## Periodicity Weekly Historical Periods 28 Forecast Periods 28 Open the Manage Forecast Models app.
Choose Create Model, and choose the planning area ZSAP6.
Enter the model name PickBestModel##.
Set the periodicity to Weekly, if not already set.
Set the Historical periods value to 28.
Set the Forecast periods values to 28.
In the Forecasting Steps tab, add the following algorithms to the forecast model: Single Exponential Smoothing, Automated Exponential Smoothing, and Simple Average. Add the forecasting step information from the table, if applicable. In the Post-processing Steps tab, select MAPE as the forecast error calculation method.
Field Value Single Exponential Smoothing Alpha Coefficient 0.3 Automated Exponential Smoothing Scope of Optimization Select Smoothing Algorithm with Best Results Measure Used for Optimization MAPE Periods in a Season 12 Limit for Alpha 0.2 – 0.6 (Lower Limit - Upper Limit) Limit for Beta 0.2 – 0.6 (Lower Limit - Upper Limit) Limit for Gamma 0.2 – 0.6 (Lower Limit - Upper Limit) Field Value Utilize Multiple Forecasts Choose Best Forecast Measure Used for Forecast Comparison MAPE Test Phase Periods 3 Choose the Forecasting Steps tab.
Set the main input key figure to Delivered Qty Adjusted, and the target key figure to Statistical Forecast Qty.
In the Save Ex-Post Forecast in, choose Expost Forecast Qty.
Choose the + sign to add the Single Exponential Smoothing algorithm.
Confirm that the alpha value is set as indicated in the table.
Choose the+ sign to add the Automated Exponential Smoothing algorithm with the settings provided in the table provided.
Choose the + sign to add the Simple Average algorithm. You do not need to maintain any further settings for this algorithm.
In the Utilize Multiple Forecasts section, choose the values given in the second table provided.
Go to the Post-Processing Steps tab.
Choose the target key figure for MAPE: Model Fit Error: MAPE.
Choose Save.
Task 2: Execute Statistical Forecasting in the Add-in for Microsoft Excel
The add-in for Microsoft Excel is used to execute the statistical forecast. In this exercise, we run the statistical forecast and view the results. Since the forecast model was created using the SAP Fiori app, it is necessary to log off and log back on to the add-in for Microsoft Excel to synchronize the objects.
Steps
Run your PickBestModel## statistical model in your Demand Planning favorite for all of your Product IDs ending in your ## at the Product ID, Location ID, Customer ID level and view the results.
Field Value Time Period Weekly Planning Level Product ID, Location ID, Customer ID Forecast model to be used PickBestModel## Target UoM PC Scenario Baseline Filter Prod ## In the Application Jobs part of the ribbon, choose Statistical Forecasting and choose Run.
A message will pop up informing you that the settings of the view will be copied over to the job prompt. Choose OK.
Set the Time Period to Weekly.
Confirm that the Planning Level is defined by Product ID, Location ID, and Customer ID.
In the Forecast model to be used field, choose the PickBestModel##, which was just created.
Ensure that the Baseline scenario is selected.
Choose PC for the UOM To ID value.
In the Filter tab, confirm that your filter Prod ## is entered.
Choose Next.
Choose Run (the Reason Code and Comment are optional).
A message will pop up informing you about the status of the job with the option to navigate to the status. Choose Navigate to Status to see the status of the job or choose OK to close the pop up.
Track the status of the execution of the forecasting job in the log. Wait until it completes.
After the job has finished, choose Refresh in the Data Input section of the ribbon bar and view the forecast result in the planning view of the Microsoft Excel add-in.
Determine which algorithms were selected for each Product, Location, Customer combination.
Choose Statistical Forecasting → Status in the Application Jobs subsection.
Locate your forecasting job in the list and choose Show Business Log.
Observe the results provided by the system and see how the system chose the model.
An analysis of the business log shows the values in the test phase and how each algorithm was selected.
For each Product ID, Location ID, and Customer ID combination, list the forecast algorithm that has been selected along with the MAPE.