NOTE: Forecast150 is released in v15, and the Forecast special expression is decommissioned as of v15.0.1.
Prior versions of Oracle Retail Demand Forecasting (RDF) use only the Forecast procedure both to generate the forecast and to estimate the promotion effects. RDF v15 introduces two procedures:
- Forecast150 procedure — generates the forecast for all forecast methods (the focus of this blog)
- CausalEstimate procedure — calculates the promotion effects at both final or source level.
While the main purpose of the Forecast150 procedure is to support aggregate causal modeling, there are additional advantages over the previous special expression, Forecast, which are described in the v15 Release Notes:
The new forecast special expression is more flexible. Building and maintaining rules is easier because it accepts the input parameter measures at different intersections as long as the input can be mapped to the forecast level. For instance, for an item/store/week forecast level, the new forecast expression can accept the forecast start date as scalar, at class/region or department level.
The new special expression is coded in a new high performance I/O caching framework, which leads to better performance.
In 15.0, the causal effects estimation is split from forecasting. This modularization allows users to specify only the meaningful inputs for specific steps in the forecasting process. This simplifies the RPAS rules and makes them easier to write and maintain.
The new forecast expression together with the new causal estimation special expression are the backbone of the Aggregated Causal Modeling approach discussed previously. This has been proven to improve the promotional forecast accuracy and robustness.
Customers upgrading to v15.0.1 should take particular note of the new expression with its new signature and plan to update their configuration accordingly.
Take advantage of the new and improved forecasting special expression, Forecast150, today!