Data Requirements (Optimization - Multifactor Elasticity)
- Ondřej Tesař
A Multifactor Elasticity model needs a Transactions Datamart to run. Here are the prerequisites for the fields.
One source is needed, it can be either a Datamart or a Data Source. It can have mixed types.
The source should contain a minimum of 2 years of transaction history, with the following columns:
Field | Required? | Comment |
|---|---|---|
Date | Yes |
|
Product | Yes | Typically Product ID. |
Quantity Measure | Yes | It is better to avoid negative or null values (they are filtered out by default by the model itself). It is possible to run the model with the log of the quantity. This may produce better outputs when there are small and large quantities in the dataset (e.g. many long-tail products). |
Shelf Price (extended) | Yes | Typically end customer price x quantity Extended to the quantity. It is better to avoid negative or null values. |
Store | Yes | Typically the store, but it can be channel or sales org. If no such feature exists, just provide a field with always the same value. |
List price (extended) | No | Extended to the quantity. List price, recommended selling price or reference price. This is used to calculate discount rate (discountPct) based on the Shelf Price. It is highly recommended to include this where possible. |
Additional categorical features | No | Here you may include any categorical columns that may influence sales. While not strictly required to run the model, this is key to getting proper outputs. Examples include Product Category, Competitor Name, Season, Channel, Special Event, Product Life cycle, Tag for promotions/discounts. |
Additional numerical features | No | Numerical features may include any numerical attributes that can be averaged over the selected time period. |
It is also required to set the data filter in “Filter” to get complete periods (and not e.g. half week at the beginning or at the end of the scope).
