System and method for generating demand groups

Patent metadata

Field	Value
Publication number	US-9785953-B2
Application number	US-72066110-A
Country	US
Kind code	B2
Filing date	Mar 9, 2010
Priority date	Dec 20, 2000
Publication date	Oct 10, 2017
Grant date	Oct 10, 2017

Abstract

The present invention relates to a system and method for generating demand groups. The system receives demand group modeling data including a product listing, point of sales data, available econometric data and product information. Attributes may then be assigned to the products based upon product identifiers, size, flavor, brand, and product descriptions utilizing natural language processing. The products may then be clustered according to the attributes and point of sales data. One or more decision trees may be generated for the product listings using the point of sales data. Demand rules may be received, which may be applied to the product clusters and the decision trees to generate demand groups. A confidence score may be generated for each product indicating how well that product fits within the demand group. These confidence scores may be compared against a threshold. Products with scores below the threshold may be flagged for user review.

First claim

What is claimed is:

1. A computer implemented method for generation of demand groups comprising:
receiving a product listing via a processor, wherein the product listing includes products comprising existing products of a product line and at least one new product with insufficient information for demand modeling that is being considered for inclusion in the product line;
receiving product data for each of the products in the product listing and demand coefficients for existing products in the product listing via the processor;
scanning the existing products via scanners at a network node to generate point of sales data and receiving the point of sales data for the existing products in the product listing over a network via the processor;
assigning attributes to each of the products of the product listing based upon the product data via the processor, wherein the product data includes product descriptors, and assigning the attributes to the products includes natural language processing of the product descriptors via the processor, wherein the product descriptors are provided in a natural language and the natural language processing enables the processor to generate machine readable attribute data and indicate meanings by annotating terms of the product descriptors with information indicating corresponding characteristics to assign the attributes;
analyzing, via the processor, the point of sales data and generating a transition matrix including one or more pairs of existing products and, for each pair of products, a sum of a quantity of occurrences of transitions within a series of purchases to identify substitutable products, wherein a transition is represented by a switch from a first product of that pair in a first purchase to a second product of that pair in a second immediately succeeding purchase;
clustering data of the existing products and at the least one new product of the product listing via the processor by:
clustering data of the existing products based on the received point of sales data and demand coefficients, the transition matrix, and the assigned attributes including the machine readable attribute data and meanings via the processor; and
clustering data of the at least one new product by:
determining a distance between the clustered data of the existing products and the assigned attributes of the at least one new product including the machine readable attribute data and meanings from the natural language processing of the product descriptors; and
clustering the data of the at least one new product with the clustered data of the existing products based on the determined distance;
generating a decision tree modeling consumer decisions for existing products of the product listing utilizing the point of sales data via the processor;
receiving, via the processor, at least one demand rule specifying one or more criteria for the demand groups;
generating demand groups of the substitutable products, via the processor, by applying the received at least one demand rule to at least one of the decision tree and the clusters of data of the products of the product listing; and
setting prices based on the generated demand groups.

2. The method for generating the demand groups, as recited in claim 1 , further comprising receiving a confidence threshold for the generation of the demand groups.

3. The method for generating the demand groups, as recited in claim 2 , further comprising generating a confidence level for each product, wherein the confidence level indicates accuracy of the placement of the product in the demand group.

4. The method for generating the demand groups, as recited in claim 3 , further comprising comparing the confidence level of each product to the confidence threshold, and when the confidence level is below the confidence threshold flagging the product for user review.

5. The method for generating the demand groups, as recited in claim 1 , wherein the clustering data of the existing products utilizes at least one of hierarchical clustering, k-means clustering, locality sensitive hashing, QT clustering, EM algorithms and model based clustering.

6. The method for generating the demand groups, as recited in claim 5 , further comprising assigning weights to the assigned attributes, wherein the weights are used by the clustering of data of the products of the product listing.

7. The method for generating the demand groups, as recited in claim 6 , further comprising updating the weights of the assigned attributes when new product data becomes available.

8. The method for generating the demand groups, as recited in claim 7 , wherein the updated weights of the assigned attributes are set between old weights and weights of the new product data.

9. The method for generating the demand groups, as recited in claim 1 , wherein the at least one demand rule includes at least one of a prioritization of size, type and brand for a division of products into the demand groups.

10. The method for generating the demand groups, as recited in claim 1 , wherein generating demand groups using decision trees includes measuring distance of products to one another in the decision tree.

11. The method for generating the demand groups, as recited in claim 1 , wherein utilizing the generated demand groups comprises:
reformatting demand group information by aggregating elements of the demand group information and processing the aggregated elements together as a single element to reduce a processing time of the processor.

12. A system for generation of demand groups comprising:
one or more network nodes including scanners to generate point of sales data;
at least one processor coupled to the one or more network nodes over a network and including:
an econometric coefficient analyzer configured to receive a product listing, demand coefficients, and point of sales data, wherein the product listing includes products comprising existing products of a product line and at least one new product with insufficient information for demand modeling that is being considered for inclusion in the product line, wherein the point of sales data and demand coefficients are received for the existing products and the point of sales data is received over a network from the one or more network nodes, and to analyze the point of sales data and generate a transition matrix including one or more pairs of existing products and, for each pair of products, a sum of a quantity of occurrences of transitions within a series of purchases to identify substitutable products, wherein a transition is represented by a switch from a first product of that pair in a first purchase to a second product of that pair in a second immediately succeeding purchase;
an attribute engine configured to receive product data for each of the products in the product listing, and further configured to assign attributes to each of the products of the product listing based upon the product data, wherein the product data includes product descriptors, and assigning the attributes to the products includes natural language processing of the product descriptors via the at least one processor, wherein the product descriptors are provided in a natural language and the natural language processing enables the at least one processor to generate machine readable attribute data and indicate meanings by annotating terms of the product descriptors with information indicating corresponding characteristics to assign the attributes;
a clustering engine configured to cluster data of the existing products and the at least one new product of the product listing by:
clustering data of the existing products based on the received point of sales data and deman

Assignees

• Desal Paritosh
• Gajendran Kamal
• IBM

Inventors

• Desal Paritosh
• Gajendran Kamal

Classifications

• G06Q30/02 Primary

  Marketing; Price estimation or determination; Fundraising · CPC title

• G06Q30/0205

  based on location or geographical consideration · CPC title

• G06Q30/0202

  Market predictions or forecasting for commercial activities · CPC title