Computer-implemented systems and methods for product attribute estimations

• US 8,065,203 B1
• Filed: 12/21/2007
• Issued: 11/22/2011
• Est. Priority Date: 08/01/2007
• Status: Active Grant

• Current Assignee: SAS Institute Incorporated
• Original Assignee: SAS Institute Incorporated
• Inventors: Joshi, Mahesh V., McGuirk, Ann Mary, Chien, Yung-Hsin
• Primary Examiner(s): Champagne; Luna
• Application Number:
  US12/004,554
• Time in Patent Office: 1,432 Days
• Field of Search: None
• US Class Current: 705/28
• CPC Class Codes: G06Q 10/06315 Needs-based resource requirements planning or analysis
• CPC Class Codes: G06Q 10/087 Inventory or stock management, e.g. order filling, procurement or balancing against orders
• CPC Class Codes: G06Q 20/203 Inventory monitoring
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities

Abstract

Systems and methods for providing estimations for a product for purchase at a plurality of stores. Groups of stores are generated based upon similarity of store demand data. For each group, a distribution is determined with respect to the attribute of the product. The distribution is used to provide estimations with respect to the product to be provided at the stores.

Claims

1. A computer-implemented method for estimating demand for a product at a plurality of stores, wherein the product is sold in several varieties, comprising:

receiving, using one or more data processors, a selection of a first attribute, wherein the first attribute is used to divide the plurality of stores into a plurality of groups;

generating, using the one or more data processors, a product hierarchy for each group, wherein the product hierarchy includes a plurality of levels containing a plurality of nodes arranged in parent-child relationships, wherein each parent node includes a total of each variety of the product sold at associated child nodes;

generating, using the one or more data processors, an aggregate distribution of demand at each parent node based upon the total of each variety of the product sold at the associated child nodes;

evaluating, using the one or more data processors, each node to identify robust nodes, wherein a robust node is identified when the aggregate distribution of a child node differs from the aggregate distribution of a parent node by more than a prespecified amount;

determining, using the one or more data processors, an aggregate distribution for each child node based on a parent node, wherein each child node inherits the aggregate distribution of a closest parent node that is a robust node; and

using the aggregate distributions at each child node to estimate demand for each variety of the product using the one or more data processors, wherein demand at each child node is estimated based upon the aggregate distribution of a nearest robust parent node at a higher level of the product hierarchy.

2. The method of claim 1, further comprising:

receiving configuration data used to remove one or more stores from the plurality of stores that are divided into the plurality of groups.

3. The method of claim 2, wherein the configuration data includes store characteristic data specifying a characteristic to use to remove one or more stores from the plurality of stores.

4. The method of claim 3, wherein the characteristic is selected from the group consisting of:

a store being a non-comparable store, a store being a new store, and a store having a low volume of historical data.

5. The method of claim 1, further comprising:

receiving, using the one or more data processors, store attribute data for a new store, wherein the new store is not in one of the plurality of groups; and

assigning, using the one or more data processors, the new store to one of the groups based upon the store attribute data.

6. The method of claim 1, wherein each store has corresponding store sales data, wherein the store sales data is adjusted for a particular store for periods when demand was constrained, and wherein the adjusted store sales data is the first attribute used to divide the plurality of stores.

7. The method of claim 6, wherein the store sales data is point-of-sale data.

8. The method of claim 6, further comprising:

imputing, using the one or more data processors, additional sales data to the particular store for periods when demand was constrained, wherein the additional sales data includes supplemental sales data from a different store having a similar sales pattern.

9. The method of claim 8, wherein the constrained demand includes an out-of-stock condition.

10. The method of claim 1, wherein each store has corresponding store demand data, wherein the store demand data is cleansed, and wherein the cleansed store demand data is the first attribute used to divide the plurality of stores.

11. The method of claim 10, wherein the data cleansing includes cleansing the store demand data by comparing sales and inventory figures, and wherein the comparison is used to determine whether the store demand data should be adjusted.

12. The method of claim 1, wherein multiple child nodes of a particular parent node inherit a particular non-zero aggregate distribution from the particular parent.

13. The method of claim 1, wherein an aggregate distribution represents portions of total sales for the product that are represented by variations of the product.

14. The method of claim 1, wherein the product is a clothing product, wherein the aggregate distribution identifies sales of small, medium, and large variations of the clothing product.

15. The method of claim 1, wherein demand is estimated for each variety of the product by generating a forecast at each child node based on the aggregate distributions at each child node.

16. A computer-implemented system for estimating demand of a product at a plurality of stores, wherein the product is sold in several varieties, the system comprising:

one or more processors;

a computer-readable storage medium containing instructions to cause the one or more processors to perform operations, including;

receiving a selection of a first attribute, wherein the first attribute is used to divide the plurality of stores into a plurality of groups;

generating a product hierarchy for each group, wherein the product hierarchy includes a plurality of levels containing a plurality of nodes arranged in parent-child relationships, wherein each parent node includes a total of each variety of the product sold at associated child nodes;

generating an aggregate distribution of demand at each parent node based upon the total of each variety of the product sold at the associated child nodes;

evaluating each node to identify robust nodes, wherein a robust node is identified when the aggregate distribution of a child node differs from the aggregate distribution of a parent node by more than a prespecified amount;

determining an aggregate distribution for each child node based on a parent node, wherein each child node inherits the aggregate distribution of a closest parent node that is a robust node; and

using the aggregate distributions at each child node to estimate demand for each variety of the product, wherein demand at each child node is estimated based upon the aggregate distribution of a nearest robust node at a higher level of the product hierarchy.

17. A computer program product for estimating demand for a product at a plurality of stores, wherein the product is sold in several varieties, tangibly embodied in a machine-readable storage medium, including instructions configured to cause a data processing system to:

receive a selection of a first attribute, wherein the first attribute is used to divide the plurality of stores into a plurality of groups;

generate a product hierarchy for each group, wherein the product hierarchy includes a plurality of levels containing a plurality of nodes arranged in parent-child relationships, wherein each parent node includes a total of each variety of the product sold at associated child nodes;

generate an aggregate distribution of demand at each parent node based upon the total of each variety of the product sold at the associated child nodes;

evaluate each node to identify robust nodes, wherein a robust node is identified when the aggregate distribution of a child node differs from the aggregate distribution of a parent node by more than a prespecified amount;

determine an aggregate distribution for each child node based on a parent node, wherein each child node inherits the aggregate distribution of a closest parent node that is a robust node; and

use the aggregate distributions at each child node to estimate demand for each variety of the product, wherein demand at each child node is estimated based upon the aggregate distribution of a nearest robust parent node at a higher level of the product hierarchy.