Tip: The Demantra Local Application replaces Collaborator Workbench. You may see both names in this text.
This chapter covers the following topics:
CDP Forecasting focuses on generating detailed, consumer-centered forecasts. When store-level data is available, a forecast is generated for the consumer demands of each item at a store with daily granularity. When store information is not available, a forecast is based on shipments issued from the customer's distribution center. The output of the forecasts is then used to generate a sell-in shipment forecast as well as basic replenishment orders.
Each engine profile is a combination of configurations and settings required as inputs for the Demantra Analytical Engine. Each profile is comprised of parameters, causal factors, forecasting models, and a forecast tree. Engine profiles are available to support the variety of analytical configurations needed to meet the different business processes in Demantra. The following engine profiles are available for CDP:
CDP Store Sell through Batch - Use this batch profile to generate the forecast shown in the Store Sell through Forecast Series. The profile is executed using the CDP Engine Store Consumption workflow. This Series appears in the CDP Sell-in Forecast Item/Store worksheet.
CDP Store Sell through Simulation - Use this simulation profile to generate the what-if forecast shown in Store Sell through Forecast Simulation Series. This profile is used to simulate at the store-level data. When accepting this simulation, values are incorporated into the Store Sell through Forecast Series
CDP Site Sell through Batch - Use this batch profile to generate the forecast shown in the Sell through Forecast Series. The profile is executed using the CDP Engine Org Site Consumption workflow. This Series appears in the CDP Sell-in Forecast Item/Customer DC worksheet.
CDP Site Sell through Simulation - Use this simulation profile to generate the what-if forecast shown in the Sell through Forecast Simulation Series. This Series appears in the CDP Sell-in Forecast Item/Customer DC worksheet. This profile generates a sell-through forecast at the site level and is not applicable at the store level. When accepting this simulation, values are incorporated into Sell through Forecast Series.
Refer to the "Engine Profiles" in the Oracle Demantra Implementation Guide for more information on available engine profiles and their uses.
The simulation and batch profiles that generate Sell-through forecasts at Site level are closely aligned with the Base engine profile. Differences are limited to the forecast tree, which is based on the Site and Account levels.
| Level | Item Level | Location Level |
|---|---|---|
| 1 | Lowest Item Level | Lowest Location Level |
| 2 | Item | Site |
| 3 | Item | Account |
| 4 | Product Category | Account |
| 5 | Highest Fictive Level | Highest Fictive Level |
In this forecast tree, the forecast is first attempted at the Item and Site level. Nodes not forecasted successfully go up to the Item and Account aggregation and finally to Product Category and Account.
The expression used for historical demand (QUANTITY_FORM), shown below, is set to use sell-through data including any user overrides.
greatest(nvl(cdp_sell_thru_hist_ovr,nvl(cdp_sell_thru_hist, 0)),0)
The requirements associated with the store-level forecast are dramatically different than a site-level forecast. The primary reason for this is the fact that the engine is generating a forecast on daily demand based on a completely different data table.
To accomplish this there are many differences between the forecasting profiles associated with store-level forecasting the with the Base engine profile. The difference includes the forecast tree, which is based on Store.
| Level | Item Level | Location Level |
|---|---|---|
| 1 | Lowest Item Level | Lowest Location Level |
| 2 | Item | Store |
| 3 | Product Category | Store |
| 4 | Highest Fictive Level | Highest Fictive Level |
In this forecast tree, the forecast is first attempted at the Item and Store level. Nodes not forecasted successfully go up to the Product Category and Store.
Note: The forecast tree currently does not go above the Store level.
During implementation it may be useful to review the forecast tree and consider including Store Group or Account.
Historical data and forecast generation are performed on the Consumption-Driven Planning data table and engine parameters have been set to support this.
The expression used for historical demand (QUANTITY_FORM), shown below, is set to use sell-through data including any user overrides.
nvl(cdp_st_sell_thru_hist_ovr ,nvl(cdp_st_sell_thru_hist,0))
Additional parameters tied to number of periods have been set to reflect a daily demand stream and the requirements for this type of data. Causal factor definitions include both daily and monthly seasonality. The PARAMETERS table used by the store-level profiles is based on definitions more suited to daily data.