Forecasting Tables for Feature-Based Forecasting

This topic explains how forecasting tables affect the outcome of feature-based forecasting for new products.

The forecasting table determines the aggregation level of the following:

• The input measure for the forecasting profile.

• The feature data that varies by the item level, planning time level, and combinations of the organization, customer site, and demand class levels and is extracted from Oracle Demand Management to the Oracle Internet of Things Intelligent Applications Cloud Service through the Oracle Business Intelligence Cloud Connector.

The forecasting table determines the data partitioning or modeling level, which is the level at which the following happen:

• Learning of the data for the input measure and time varying feature data of existing products.

• Generation of the forecast recommendations for the new product.

Ensure that these prerequisites are met while setting up the forecasting table:

• The time level for the forecasting table should match that of the demand plan's planning time level and planning calendar. The forecasting table can't contain multiple time levels.

  If the demand plan doesn't use a hybrid time hierarchy, select Day, Week, Gregorian Month, or fiscal Month as the time level. If the planning time level is Day, the Day level in the forecasting table doesn't need to be from the time hierarchy that you selected as the planning calendar. If the planning time level is Week, Gregorian Month, or fiscal Month, the Week, Gregorian Month, or fiscal Month level must be from the time hierarchy that you selected as the planning calendar.

  If the demand plan uses a hybrid time hierarchy, select the Week, Gregorian Month, or fiscal Month level from the hybrid time hierarchy that's selected as the planning calendar or a corresponding hybrid time hierarchy.

  Note: The forecasting time level selected on the Demand tab on the Plan Options page for the demand plan doesn't apply to feature-based forecasting.

• The Product dimension must be selected for the table.

• The forecasting table must contain the input and output measures that you plan to use for feature-based forecasting.

Data Partitioning Levels

This table lists the data partitioning levels for various combinations of levels and the planning time level:

Combination	Data Partitioning Level
Item, organization, and planning time level	Organization
Item, customer site, and planning time level	Customer site
Item, organization, customer site, and planning time level	Organization and customer site
Item, organization, demand class, and planning time level	Organization and demand class
Item, customer site, demand class, and planning time level	Customer site and demand class
Item, organization, customer site, demand class, and planning time level	Organization, customer site, and demand class