Scalable regression for retail panel data

A computing system comprising

• a database configured to store a demand model for a plurality of retail products that includes effects of base demand and seasonality, an elasticity equation comprising a transformed version of the demand model transformed such that the transformed demand model does not depend on base demand or seasonality, the database further configured to store retail panel data describing historic demand for the plurality of retail products over time;
• a processor configured to cause the computing system to operate with a group of logics, the group of logics including:
• a data transformation logic configured to transform retail panel data to cancel effects of base demand and seasonality for a given retail item;
• and
• a solution logic configured to solve the elasticity equation using the transformed retail panel data to estimate an elasticity of demand for the retail item and to estimate base demand and seasonality for the retail item based, at least in part, on the estimated elasticity of demand for the retail item and the demand model.

A non-transitory computer-readable medium storing computer-executable instructions that when executed by a computer cause the computer to perform a method, the method comprising:

• performing a regression that estimates elasticity of demand for a retail item, wherein the regression is performed on a transformation of a demand model and retail demand data that have been transformed to cancel the effects of base demand and seasonality;
• in a subsequent processing step, estimating a base demand and seasonality for the retail item based, at least in part, on the estimated elasticity of demand and an un-transformed version of the demand model that includes the effects of base demand and seasonality;
• and
• wherein the performing a regression and estimating are performed in a database that stores retail panel data for the retail item and other retail items.

A computer-implemented method comprising:

• accessing, by a processor, retail panel data describing historic demand for a plurality of retail products over time;
• transforming, by the processor, the retail panel data to cancel effects of base demand and seasonality for a first retail item;
• solving, by the processor, an elasticity equation using the transformed retail panel data to estimate an elasticity of demand for the first retail item;
• where the elasticity equation comprises a transformed demand model for the plurality of retail products that has been transformed to cancel effects of base demand and seasonality;
• estimating, by the processor, base demand and seasonality for the first retail item based, at least in part, on the estimated elasticity of demand for the first retail item and an un-transformed version of the demand model that includes effects of base demand and seasonality;
• providing, by the processor, the estimated elasticity, based demand, and seasonality for use in determining a sales forecast for the first retail item.