Measures for Feature-Based Forecasting

This topic lists the measures that are used in feature-based forecasting.

Input Measures

You can use the predefined Final Bookings History or Final Shipments History measure as the input measure for feature-based forecasting.

You can also select a user-defined measure as the input measure if it's dimensioned by the Product and Time dimensions. Additionally, the user-defined measure can be dimensioned by the Organization and Customer dimensions.

The input measure must be present in the forecasting table of the forecasting profile that's used for feature-based forecasting and in the measure catalog for the demand plan. If you specify a user-defined measure catalog for the forecasting profile, the input measure is automatically added to this measure catalog.

Output Measures

You can use the predefined Feature Based Bookings Forecast or Feature Based Shipments Forecast measure as the output measure for feature-based forecasting.

Note: Oracle recommends that you don't use the same output measure for feature-based forecasting and Bayesian forecasting.

In addition to the bookings or shipments forecast, feature-based forecasting also provides the feature importance and model accuracy, which are provided in these predefined measures:

• Feature Based Bookings Forecast: Feature Importance

• Feature Based Bookings Forecast: Model Accuracy

• Feature Based Shipments Forecast: Feature Importance

• Feature Based Shipments Forecast: Model Accuracy

If you select a user-defined output measure for feature-based forecasting, note the following:

• The output measure should have the same dimensions as the input measure for feature-based forecasting.

• Two measures are automatically created for storing the feature importance and model accuracy and named <user-defined measure>: Feature Importance and <user-defined measure>: Model Accuracy respectively. You must add these measures to the measure catalog for the demand plan.

The output measure must be present in the forecasting table of the forecasting profile that's used for feature-based forecasting and in the measure catalog for the demand plan. If you specify a user-defined measure catalog for the forecasting profile, the output measure is automatically added to this measure catalog.

Other Measures

This table lists other predefined measures for feature-based forecasting:

Predefined Measure	Description
Bookings Forecast Type	Type of the statistical method used for forecasting bookings of the new product. Used in the calculation of the Final Bookings Forecast Based on Forecast Type measure.
Feature Based Forecasting: Adjusted Bookings Forecast	Adjustment you enter for the feature-based forecast of bookings of the new product. Overrides the value of the Feature Based Bookings Forecast measure.
Final Feature Based Bookings Forecast	Final feature-based bookings forecast for the new product based on the bookings history and user adjustments. Is equal to the Feature Based Forecasting: Adjusted Bookings Forecast measure if it has a value, or is equal to the Feature Based Bookings Forecast measure.
Final Bookings Forecast Based on Forecast Type	Final bookings forecast for the new product that's based on the bookings history, bookings forecast type, and user adjustments. Is equal to the Final Bookings Forecast measure if the Bookings Forecast Type measure is "Bayesian." Is equal to the Final Feature Based Bookings Forecast measure if the Bookings Forecast Type measure is "Feature-based."
Shipments Forecast Type	Type of the statistical method used for forecasting shipments of the new product. Used in the calculation of the Final Shipments Forecast Based on Forecast Type measure.
Feature Based Forecasting: Adjusted Shipments Forecast	Adjustment you enter for the feature-based forecast of shipments of the new product. Overrides the value of the Feature Based Shipments Forecast measure.
Final Feature Based Shipments Forecast	Final feature-based shipments forecast for the new product based on the shipments history and user adjustments. Is equal to the Feature Based Forecasting: Adjusted Shipments Forecast measure if it has a value, or is equal to the Feature Based Shipments Forecast measure.
Final Shipments Forecast Based on Forecast Type	Final shipments forecast for the new product that's based on the shipments history, shipments forecast type, and user adjustments. Is equal to the Final Shipments Forecast measure if the Shipments Forecast Type measure is "Bayesian." Is equal to the Final Feature Based Shipments Forecast measure if the Shipments Forecast Type measure is "Feature-based."
Product Launch Date	Date when the product is available for sale in the market. Forecast recommendations are generated from this date. This measure is used in the Product Launch Date feature group for a forecasting profile that uses feature-based machine learning. This measure is mapped to the GLOBAL_ATTRIBUTE_DATE1 attribute for the Item level of the Product hierarchy and populated through the regular collection process or the file-based data import (FBDI) template for Supply Chain Planning measures. Deselect the Active check box for this measure if you create a user-defined date measure with the Product dimension for storing the product launch date. Create a feature group for the user-defined date measure, and either edit it or load it through the FBDI template for Supply Chain Planning measures. Tip: If you create a user-defined measure for the product launch date, you can edit the measure expression of the predefined Product Launch Date measure so that it points to the user-defined measure and keep the Active check box for the Product Launch Date measure selected. By doing so, you can avoid editing the expressions for the other predefined measures that refer to the Product Launch Date measure.
Product Launch Date: Month Number	Number of the Gregorian calendar month of the product launch. Used to analyze the impact of the launch month on product sales. This measure is used in the Product Launch Month feature group for a forecasting profile that uses feature-based machine learning. If the product launch date is stored in a user-defined measure, the expression of the Product Launch Date: Month Number measure has to be edited so that the user-defined measure is referred to. You don't need to take this action if the Product Launch Date measure refers to the user-defined measure.
Product Launch Date: Relative Day Number	Day number calculated with reference to the product launch date. Set to 1 for the day of the launch, and increased by 1 for every successive day. Internal measure used in the expression of the Product Launch Date: Relative Time Period Number measure for demand plans that are defined at the day planning time level. If the product launch date is stored in a user-defined measure, the expression of the Product Launch Date: Relative Day Number measure has to be edited so that the user-defined measure is referred to. You don't need to take this action if the Product Launch Date measure refers to the user-defined measure.
Product Launch Date: Relative Week Number	Week number calculated with reference to the product launch date. Set to 1 for the week of the launch date, and increased by 1 for every successive week. Internal measure used in the expression of the Product Launch Date: Relative Time Period Number measure for demand plans that are defined at the week planning time level. If the product launch date is stored in a user-defined measure, the expression of the Product Launch Date: Relative Week Number measure has to be edited so that the user-defined measure is referred to. You don't need to take this action if the Product Launch Date measure refers to the user-defined measure.
Product Launch Date: Relative Month Number	Number of Gregorian calendar month calculated with reference to the product launch date. Set to 1 for the month of the launch date, and increased by 1 for every successive month. Internal measure used in the expression of the Product Launch Date: Relative Time Period Number measure for demand plans that are defined at the planning time level of the Gregorian calendar month. If the product launch date is stored in a user-defined measure, the expression of the Product Launch Date: Relative Month Number measure has to be edited so that the user-defined measure is referred to. You don't need to take this action if the Product Launch Date measure refers to the user-defined measure.
Product Launch Date: Relative Time Period Number	Time period number calculated with reference to the product launch date. Set to 1 for the time period of the launch date, and increased by 1 for every successive time period. Used for determining the shape of the demand curve for the new product's launch. This measure is used in the Demand Curve Shape for Product Launch feature group for a forecasting profile that uses feature-based machine learning. The measure is configured for a demand plan with the week planning time level. If your demand plan has the day or month planning time level, you need to edit the expression of the Product Launch Date: Relative Time Period Number measure so that it refers to the Product Launch Date: Relative Day Number or Product Launch Date: Relative Month Number measure.
Seasonality	Month number according to the Gregorian calendar for the time period of the demand history. Used for analyzing the seasonality of demand data. This measure is used in the Monthly Seasonality feature group for a forecasting profile that uses feature-based machine learning.