Multi-channel demand planning for inventory planning and control

Patent metadata

Field	Value
Publication number	US-12079828-B2
Application number	US-202117529075-A
Country	US
Kind code	B2
Filing date	Nov 17, 2021
Priority date	Nov 17, 2021
Publication date	Sep 3, 2024
Grant date	Sep 3, 2024

Abstract

Methods and systems for forecasting demand for items across multiple channels are disclosed. In some implementations, multi-channel demand forecasting may be performed on a per-item, per-location basis, by selectively generating item-location forecasts for each item and location within a supply chain for each channel, or disaggregating a chain level forecast on a per-item basis to each location. Particular selection of an appropriate model, and selective training of models, allows for efficient computation of such forecasts across a large supply chain with thousands of locations and hundreds of thousands, or millions, of items for which forecasts are generated.

First claim

The invention claimed is:

1. A system for forecasting demand for each of a plurality of items sold through multiple sales channels of a retail enterprise having a plurality of locations, the system comprising:
at least one processor and at least one memory device, the memory device storing instructions that, when executed by the at least one processor, cause the system to perform:
training a generalized additive model (GAM) using historical sales data of a retail enterprise, the historical sales data including online sales data, to determine an overall item demand forecast for a future time period;
separating the overall item demand forecast into a localized item demand forecast and a non-localized item demand forecast based, at least in part, on an item eligibility for localized item fulfillment and historical localized and non-localized item ordering information;
for the non-localized item demand forecast, determine a portion of non-localized item demand able to be fulfilled from a retail location and a remaining portion of non-localized item demand to be fulfilled from a warehouse;
disaggregating the portion of non-localized item demand able to be fulfilled from a retail location at a location level to attribute sub-portions of the non-localized item demand to each of a plurality of retail locations;
disaggregating the localized item demand forecast into localized item demand at each of the plurality of retail locations;
combining the disaggregated localized item demand and the disaggregated non-localized item demand for each respective one of the plurality of retail locations to determine an overall online item demand for each of the plurality of retail locations;
in response to a request for item demand at an application programming interface (API) exposed by the system from a consumer system, the request identifying at least one item and one or more of the plurality of retail locations, returning, to the consumer system, an overall online item demand forecast for the at least one item at the one or more of the plurality of retail locations;
optimizing a plurality of models, the plurality of models including the GAM, wherein optimizing the plurality of models comprises selectively retraining the plurality of models by retraining a first one or more of the plurality of models at a first time and retraining a second one or more of the plurality of models at a second time that is delayed relative to the first time;
wherein selectively retraining the plurality of models comprises:
determining that the first one or more of the plurality of models are to be retrained at the first time based on an update to demand data for a first one or more items forecasted by the first one or more of the plurality of models; and
determining that the second one or more of the plurality of models are to be retrained at the second time that is delayed relative to the first time based on a lack of an update to demand data for a second one or more items forecasted by the second one or more of the plurality of models.

2. The system of claim 1 , wherein the historical sales data further includes in-store sales data.

3. The system of claim 1 , wherein the historical sales data includes item-level sales data from each of the plurality of retail locations.

4. The system of claim 1 , wherein the instructions further cause the system to perform, in response to the request for item demand, returning an overall item demand forecast including the overall online item demand forecast and an in-store item demand forecast for the at least one item.

5. The system of claim 4 , wherein the overall online item demand forecast and the in-store item demand forecast are returned separately from each other.

6. The system of claim 4 , wherein the system is further configured to generate, from a separate demand forecast model, a second overall item demand forecast for each of the plurality of retail locations.

7. The system of claim 6 , wherein the system is configurable to select between the overall item demand forecast and the second overall item demand forecast.

8. The system of claim 1 ,
further comprising the consumer system, wherein the consumer system comprises an inventory management application;
wherein the inventory management application is configured to receive the overall online demand forecast for the at least one item at the one or more of the plurality of retail locations; and
wherein the inventory management application is configured to, in response to receiving the overall online demand forecast, automatically generate a transfer order to route the at least one item from a first retail location of the one or more of the plurality of retail locations to a second retail location of the one or more of the plurality of retail locations.

9. The system of claim 1 , wherein the at least one item includes a class of items.

10. The system of claim 1 , wherein the generalized additive model is a generalized additive mixed model (GAMM).

11. The system of claim 1 , wherein the overall item demand forecast for an item is based, at least in part, on historical sales of the item and other items having a commonality with the item.

12. A method of forecasting sales-channel specific demand through a retail supply chain fulfilling sales transactions received via multiple channels from a common set of retail locations, the method comprising:
training a generalized additive model (GAM) using historical sales data of a retail enterprise, the historical sales data including online sales data, to determine an overall item demand forecast for a future time period;
separating the overall item demand forecast into a localized item demand forecast and a non-localized item demand forecast based, at least in part, on an item eligibility for localized item fulfillment and historical localized and non-localized item ordering information;
for the non-localized item demand forecast, determine a portion of non-localized item demand able to be fulfilled from a retail location and a remaining portion of non-localized item demand to be fulfilled from a warehouse;
disaggregating the portion of non-localized item demand able to be fulfilled from a retail location at a location level to attribute sub-portions of the non-localized item demand to each of a plurality of retail locations;
disaggregating the localized item demand forecast into localized item demand at each of the plurality of retail locations;
combining the disaggregated localized item demand and the disaggregated non-localized item demand for each respective one of the plurality of retail locations to determine an overall online item demand for each of the plurality of retail locations; and
in response to a request for item demand at an application programming interface (API) exposed by the system from a consumer system, the request identifying at least one item and one or more of the plurality of retail locations, returning, to the consumer system, an overall online item demand forecast for the at least one item at the one or more of the plurality of retail locations;
optimizing a plurality of models, the plurality of models including the GAM, wherein optimizing the plurality of models comprises selectively retraining the plurality of models by retraining a first one or more of the plurality of models at a first time and retraining a second one or more of the plurality of models at a second time that is delayed relative to the first time;
wherein selectively retraining the plurality of models comprises:
determining that the first one or more of the plurality of models are to be retrained at the first time based on an update to demand data for a first one or more

Assignees

• Target Brands Inc

Inventors

• Jabri Omar N
• Riggall Adam
• Sizemore Duane
• Kim Peter
• Jelvis Tikhon
• Liu Yang
• Bhattacharya Saibal
• Majumdar Sayon
• Baz Zeynep Erkin

Classifications

• G06Q10/067

  Enterprise or organisation modelling · CPC title

• G06Q10/087

  Inventory or stock management, e.g. order filling, procurement or balancing against orders · CPC title

• G06Q30/0625

  by formulating product or service queries, e.g. using keywords or predefined options · CPC title

• G06Q10/06315

  Needs-based resource requirements planning or analysis · CPC title

• G06Q30/0202 Primary

  Market predictions or forecasting for commercial activities · CPC title