Measures for Hyperparameter Tuning

This topic lists the measures that are provided or created for use with the hyperparameter-tuning feature for demand or demand and supply plans.

If you select the predefined Bookings Forecast or Shipments Forecast measure as the output measure for the forecasting profile for hyperparameter tuning, then the corresponding measures are added the measure catalog for your plan. These measures are in the measure group named Hyperparameter-Based Forecasting.

If you select a user-defined output measure for the forecasting profile for hyperparameter tuning, then the corresponding measures are created when you save the forecasting profile. The measures are added to the measure catalog for the plan when it's run with current data, and if the plan's measure catalog has been selected for the forecasting profile. The names of these measures begin with the name of your user-defined output measure.

This table lists the measures for hyperparameter tuning:

Measure Name	Measure Description
Bookings Forecast: Tuned Shipments Forecast: Tuned <user-defined output measure>: Tuned	Provides the forecast for a forecasting node (combination) according to the tuned settings for forecasting parameters for that forecasting node.
Bookings Forecast: Tuned Base MAPE Shipments Forecast: Tuned Base MAPE <user-defined output measure>: Tuned Base MAPE	Provides the mean absolute percentage error (MAPE) for a forecasting node's forecast that's according to the forecasting parameters for the forecasting profile. This forecast is calculated on the historical demand data before the plan start date. The length of the historical demand data for holdback validation (out-of-sample testing) is specified by the HypertuneHoldback and HypertuneHoldbackPercent forecasting parameters. The percentage of the forecasting nodes that's considered in the historical demand data is specified by the HypertuneSamplePercent forecasting parameter. For a plan run, the same forecasting nodes are considered for calculation of the base MAPE and best MAPE measures.
Bookings Forecast: Tuned Best MAPE Shipment Forecast: Tuned Best MAPE <user-defined output measure>: Tuned Best MAPE	Provides the MAPE for a forecasting node's forecast that's according to the tuned forecasting parameters. This forecast is calculated on the historical demand data before the plan start date. The length of the historical demand data for holdback validation (out-of-sample testing) is specified by the HypertuneHoldback and HypertuneHoldbackPercent forecasting parameters. The percentage of the forecasting nodes that's considered in the historical demand data is specified by the HypertuneSamplePercent forecasting parameter. For a plan run, the same forecasting nodes are considered for calculation of the base MAPE and best MAPE measures.
Bookings Forecast: Tuned Settings Shipments Forecast: Tuned Settings <user-defined output measure>: Tuned Settings	Provides the following for each tuned forecasting node: Names and optimized values for the tuned forecasting parameters Status of the hyperparameter tuning for the forecasting node Level of the forecasting node in the forecast tree for the forecasting profile For more information about this measure, see the topic titled Text in the Tuned-Settings Measure for Hyperparameter Tuning.
Bookings Forecast: Last Attempt at Tuning Shipments Forecast: Last Attempt at Tuning <user-defined output measure>: Last Attempt at Tuning	Provides the date and time when hyperparameter tuning was last performed for a forecasting node whether the tuning was successful or not. For forecasting nodes that aren't included in the plan population specified by the HypertuneSamplePercent forecasting parameter, the value is 01/01/2000.