Demand Forecasting

• 15 minutes

Demand forecasting is used to predict independent demand from sales orders and dependent demand at any decoupling point for customer orders. The enhanced demand forecast reduction rules provide an ideal solution for mass customization.

To generate the baseline forecast, a summary of historical transactions is passed to a Microsoft Azure Machine Learning service. Because this service isn't shared among users, it can be customized to meet industry-specific requirements. You can use Supply Chain Management to visualize the forecast, adjust the forecast, and view key performance indicators (KPIs) about forecast accuracy.

Main features of demand forecasting include:

• Generating a statistical baseline forecast that is based on historical data.

• Using a dynamic set of forecast dimensions.

• Visualizing demand trends, confidence intervals, and adjustments of the forecast.

• Authorizing the adjusted forecast to be used in planning processes.

• Removing outliers.

• Creating measurements of forecast accuracy.

The setup tasks include setting up the following data and parameters.

Item allocation key

A demand forecast is calculated for an item and its dimensions only if the item is part of an item allocation key. This rule is enforced to group large numbers of items so that demand forecasts can be created faster. The item allocation key percentage is ignored when demand forecasts are generated. Forecasts are created based on historical data only.

An item and its dimensions must be part of only one item allocation key if the item allocation key is used during forecast creation.

To add a stock keeping unit (SKU) to an item allocation key, go to Master planning > Setup > Demand forecasting > Item allocation keys. Use the Assign items page to assign an item to an allocation key.

Intercompany planning groups

Demand forecasting generates cross-company forecasts. In Supply Chain Management, companies that are planned together are grouped into one intercompany planning group. To specify, for each company, which item allocation keys should be considered for demand forecasting, associate an item allocation key with the intercompany planning group member by going to Master planning > Setup > Demand forecasting > Intercompany planning groups.

By default, if no item allocation keys are assigned to intercompany planning group members, a demand forecast is calculated for all items that are assigned to all item allocation keys from all Supply Chain Management companies. Additional filtering options for companies and item allocation keys are available on the Generate statistical baseline forecast page.

Make sure that you review the number of items that are forecasted. Unnecessary items might cause increased costs when you use Azure Machine Learning.

Demand forecasting parameters

To set up demand forecasting parameters, go to Master planning > Setup > Demand forecasting > Demand forecasting parameters. Because demand forecasting runs cross-company, the setup is global. In other words, the setup applies to all companies.

Demand forecasting generates the forecast in quantities. Therefore, the unit of measure that the quantity should be expressed in must be specified in the Demand forecast unit field.

The unit of measure must be unique to help guarantee that the aggregation and percentage distribution make sense. For more information about aggregation and percentage distribution, see Make manual adjustments to the baseline forecast.

For every unit of measure that is used for SKUs that are included in demand forecasting, make sure that there is a conversion rule for the unit of measure and the general forecasting unit of measure. When forecast generation is run, the list of items that don't have a unit of measure conversion is logged so that you can correct the setup.

Demand forecasting can be used to forecast both dependent and independent demand.

For example, if only the Sales order check box is selected, and if all the items that are considered for demand forecasting are items that are sold, the system calculates independent demand. However, critical subcomponents can be added to item allocation keys and included in demand forecasting. In this case, if the Production line check box is selected, a dependent forecast is calculated.

Supply Chain Management provides two methods for creating a baseline forecast. You can use forecasting models on top of historical data, or you can copy the historical data to the forecast. The Forecast generation strategy field lets you select between these two methods. To use forecast models, select Azure Machine Learning.

By selecting Forecast dimensions in the left pane of the Demand forecasting parameters page, you can also select the set of forecast dimensions to use when the demand forecast is generated. A forecast dimension indicates the level of detail that the forecast is defined for. Company, site, and item allocation key are mandatory forecast dimensions, but you can also generate forecasts at the warehouse, inventory status, customer group, customer account, country and/or region, state, and item plus all item dimension levels.

At any point, you can add forecast dimensions to the list of dimensions that are used for demand forecasting. You can also remove forecast dimensions from the list. However, manual adjustments are lost if you add or remove a forecast dimension.

Not all items behave in the same manner from a demand forecasting perspective. Similar items can be grouped in one item allocation key, and parameters such as transaction types and forecast method settings can be set for each item allocation key.

Demonstration - Create a Baseline forecast

First you must set up an item allocation key:

1. Go to Master planning > Setup > Intercompany planning groups.
2. Use the Quick Filter to find records. For example, filter on the Name field with a value of 10.
3. Demand forecasting runs across legal entities. That's why you need to set up all the companies for which you want to generate forecasts in one intercompany planning group.
4. In the list, find and select the desired record.
5. Select Item allocation keys.
6. Select all the item allocation keys for which you want to create forecasts.
7. In the list, mark the selected row.
8. Select D_Aloc item allocation key.
9. Select >.
10. Close the pages

Now you will set up the demand forecasting parameters:

1. Go to Master planning > Setup > Demand forecasting > Demand forecasting parameters.
2. Expand the Forecast algorithm parameters section.
3. In the Forecast generation strategy field, select Copy over historical demand.
4. Select Save.

Now you are ready to create a baseline forecast:

1. Go to Master planning > Forecasting > Demand forecasting > Generate statistical baseline forecast.
2. In the Historical Horizon section, in the From date field, enter a date.
3. If you have sales orders starting from January 1, 2015, enter this date. If you don't, enter the earliest date of your sales orders.
4. In the To date field, enter a date.
5. Enter the last date of your sales orders, for example 2015-03-31.
6. In the Baseline Forecast Start Date section, in the From date field, enter a date.
7. Enter 2015-04-01. This date will be automatically calculated as the start date of the next forecasting bucket.
8. Expand the Records to include section.
9. Select Filter.
10. In the list, mark the selected row.
11. Mark the row where Field = Intercompany planning group.
12. In the Criteria field, type a value.
13. Type the intercompany planning group (for example, 10) that you used in the first task.
14. In the list, find and select the desired record.
15. Select the row where Field = Item allocation key.
16. In the Criteria field, type a value.
17. Select OK.
18. Expand the Advanced parameters section.
19. In the Forecast bucket field, select Month.
20. In the Forecast horizon field, enter 3.
21. In the Freeze time fence field, enter 1.
22. Select OK.

To visualize the demand forecast, follow these steps:

1. Go to Master planning > Forecasting > Demand forecasting > Adjusted demand forecast.
2. In the aggregated view table, select the cell in row 1, column 2. This is the second month in which you have created a forecast.
3. Set QtyCell to 400. a. In the cell, enter a different number than the one that was forecasted, for example, 400.
4. You have made a manual adjustment to the forecast. Notice the graphical indication in the next step.
5. Select Forecast line details. a. In this page, you can see the accuracy values, historical demand, and forecast. You can make changes to the forecast as well.

For more information regarding generating statistical baseline forecasts, see Generate a statistical baseline forecast.

Azure Machine Learning service

To generate the forecast, Supply Chain Management also uses a Machine Learning web service. To connect to the service, you must provide the following information if you sign into Azure Machine Learning studio:

• Web service application programming interface (API) key

• Web service endpoint URL

• Azure storage account name

• Azure storage account key

To create demand predictions, you can deploy your own service by using Machine Learning Studio or the Supply Chain Management demand forecasting experiments.

Instructions for deploying the Supply Chain Management demand forecasting experiments as a web service are available in Supply Chain Management. To access these instructions, on the Demand forecasting parameters page, select the Azure Machine Learning tab.

Settings for the Demand forecasting Machine Learning service

To view the parameters that can be configured for the Demand forecasting service, go to Master Planning > Setup > Demand forecasting > Forecasting algorithm parameters. The Forecasting algorithm parameters page shows the default values for the parameters.

You can overwrite these parameters on the Demand forecasting parameters page. Use the General tab to overwrite the parameters globally or use the Item allocation keys tab to overwrite the parameters for each item allocation key. Parameters that are overwritten for an item allocation key affect only the forecast of items that are associated with that item allocation key.

For more information regarding Demand forecasting overview, see Demand forecasting overview.