How You Enable Hyperparameter Tuning for a Plan

This topic provides the high-level process for enabling hyperparameter tuning for a demand or demand and supply plan.

Follow these steps to set up hyperparameter tuning for your plan:

1. Create or modify a user-defined forecasting profile of the Bayesian machine-learning type.
   Note: To add forecasting parameters for hyperparameter tuning to a predefined forecasting profile, you must make a copy of it and edit the copy.
2. In the Parameters step in the guided process for creating or modifying the forecasting profile, according to your requirement, add the forecasting parameters from the Hyperparameter Tuning and Nodal Tuning categories to the forecasting profile. For information, see the topic titled Forecasting Parameters for Hyperparameter Tuning.

   You must add the HypertuneSamplePercent forecasting parameter to your forecasting profile. If you don't set this forecasting parameter to a value greater than 0 (zero), you can set the value at the level of the plan.

   Even if you don't add the remaining forecasting parameters for hyperparameter tuning to your forecasting profile, they will apply with the default values.

3. In the Parameters step, for the forecasting parameters of other categories, in the Hyperparameter Tuning Values column, enter your values for hyperparameter tuning.

   Separate the values in the column with commas.

   If you enter values incorrectly, hyperparameter tuning won't happen for forecasting nodes (combinations), and a message to this effect will be provided through the Bookings Forecast: Tuned Settings, Shipments Forecast: Tuned Settings, or <user-defined output measure>: Tuned Settings measure.

   If you don't know the values that you need to enter, skip this step. The default set of values for hyperparameter tuning in the HypertuneParamSet forecasting parameter will be evaluated for selection of the tuned values.

4. Add the forecasting profile to a new plan or an existing one.

   On the Forecast Settings tab in the Demand step in the guided process for creating or editing your plan, add the forecasting profile.

   You can optionally change the values for some forecasting parameters at the level of the plan in the section titled Hyperparameter tuning settings in the drawer for adding or editing the forecasting profile in the guided process for editing the plan. These edited values for forecasting parameters will be saved for the plan and won't apply to the corresponding forecasting parameters in the forecasting profile. For information, see the topic titled Specify Hyperparameter-Tuning Values at the Plan Level.

5. Run the plan.

   In the Run plan drawer, in the section titled Data Refresh Options, select the Refresh with current data option.

   In the section titled Forecasting Profiles, select the forecasting profile for hyperparameter tuning.

   Note: For hyperparameter tuning to be set up for this plan and forecasting profile, you must select the Refresh with current data option.
6. Once the plan run completes, on the More Actions menu for your plan, if the Planning Advisor menu item is available, select it to open the Planning Advisor.

   The menu option is available if you've enabled the HypertuneAdvisor forecasting parameter, and if there are notifications of the Summary of hyperparameter tuning type for your plan. The number of notifications is indicated in parentheses after the menu item. For details, see the topic titled How You Set Up Hyperparameter-Tuning Notifications in the Planning Advisor.

7. Select the notification to view the summary of hyperparameter tuning. For details, see the topics titled View Hyperparameter-Tuning Notifications in the Planning Advisor and Hyperparameter-Tuning Details in the Planning Advisor.
8. Select the View More button to open a user-defined page on a tab for the plan.

   If you've configured the HypertuneAdvisorPageLayout forecasting parameter, the button will be available. You should have created the page with tables and graphs for analyzing the results of hyperparameter tuning.

9. Review the results of hyperparameter tuning.

   The tuned-settings measure for your forecasting profile will indicate the status of tuning for each forecasting node (combination) and the names and optimized values for the tuned forecasting parameters. For information, see the topic titled Text in the Tuned-Settings Measure for Hyperparameter Tuning.

   If you've set the value of the HypertuneOutputMode forecasting parameter to 1, the tuned forecast is stored in the output measure for your forecasting profile if the value you specified for the HypertuneMAPEThreshold forecasting parameter is equalled or exceeded.

   If you've set the value of the HypertuneOutputMode forecasting parameter to 2, the tuned forecast is stored in the Bookings Forecast: Tuned, Shipments Forecast: Tuned, or <user-defined output measure>: Tuned measure if the value you specified for the HypertuneMAPEThreshold forecasting parameter is equalled or exceeded. For each successfully tuned forecasting node, you can compare the value of this measure to the value of the forecasting profile's output measure. The default value for the HypertuneOutputMode forecasting parameter is 2 so that you can compare and evaluate the results of hyperparameter tuning.

   If you've set the value of the EnableNodalTuning forecasting parameter to 2 and the value of the HypertuneOutputMode forecasting parameter to 1, the tuned settings for the successfully tuned forecasting nodes in the previous plan runs are automatically applied to the same forecasting nodes during the present plan run. For information, see the topic titled How You Enable Nodal Tuning for Forecasting Nodes.