Causal forecasting

[Emerson Rush]

This series is intended as an introduction to SAP Forecasting and Replenishment, to serve as reference for subsequent discussions and troubleshooting throughout the Forecasting and Replenishment forum.

SAP Forecasting and Replenishment allows retail businesses to sustain profitability in an increasingly competitive environment, via causal-based forecasting and requirements quantity optimization that is integrated with SAP promotions, transportation and warehousing solutions. SAP F&R represents streamlined supply chain operations that translate to customer retention as demand for assortment variety and quality increases, while safeguarding low inventory and transportation costs.

At the core of SAP F&R is an array of automatically selected algorithms that support causal forecasting, developed by SAF AG and acquired by SAP. This allows for accurate replication of the effects of promotions, price changes and other Demand Influencing Factors individually for products across each retail location and distribution center.

Highly accurate sales forecasts are incorporated into optimized, supply chain-aware order quantities and order forecasts that comprise multilevel replenishment across the supply chain. SAP Forecasting and Replenishment becomes a hands off, exception-based managed solution that ensures availability while reducing inventory costs, and allows for replenishment planners to focus on their products’ lifecycle, rather than on reacting to deficiencies in order quantities

Causal Forecasting

Causal Forecasting is achieved in SAP Forecasting and Replenishment via Demand Influencing Factors, or DIF. A DIF is any event or metric that can be defined to explain – and replicate – its effect on a product’s sales. DIF can be assigned to locations (and their complete assortment of products), or to individual products, and are divided primarily into three categories: Event, Value and Correctional.

Event DIFs

Event DIFs are defined as to whether they occur or not within a specific timeframe. Some examples of event DIFs are product promotions, local occurrences of events like festivals, and temporary store closings.

Value DIFs

A DIF can be defined in reference to a value that can be related to impact on sales. An example of value DIFs could be the size of a store display on which products are advertised. Value DIFs can be treated as independent occurrences, or more interestingly, as a record of changes in values throughout time (think weather forecasts and how these affect sales for given products). These type of DIFs are known as Time Series.
There is a special DIF Type, Sales Price, which is a form of Time Series DIF pre-delivered by SAP and integrated with SAP ECC. Occurrences of this DIF are maintained once, and the analysis of price changes is incorporated into forecasting for products with frequent price changes, which don’t comprise a promotion per se.

Correctional DIFs

Finally, DIFs can be employed to arbitrarily affect forecast values, be it via a relative (percentage) or absolute correction. It is also possible to determine that certain past sales are to be excluded from the forecast calculation via the use of an Ignore DIF. This is useful, for example, when a store was closed due to remodelling or other incidents.

[Author]

Posts