Household level segmentation method and system

• US 8,364,678 B2
• Filed: 09/02/2008
• Issued: 01/29/2013
• Est. Priority Date: 05/29/2001
• Status: Expired due to Term

• Current Assignee: Claritas, LLC, The Nielsen Company (US), LLC (Nielsen Holdings plc)
• Original Assignee: The Nielsen Company LLC (MOOD Media Corporation)
• Inventors: Miller, David R., Inman, Kenneth L.
• Primary Examiner(s): Corrielus, Jean M
• Application Number:
  US12/202,805
• Publication Number:
  US 20080319834A1
• Time in Patent Office: 1,610 Days
• Field of Search: 707/736, 707/741, 707/750, 707/999.003, 707/999.005, 707/690, 707/740, 707/217, 707/713, 707/737, 704/9, 715/231, 725/35, 709/217, 705/10, 705/38, 705/4, 705/2, 705/7.34, 705/30
• US Class Current: 707/737
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 30/0204 Market segmentation
• CPC Class Codes: Y10S 707/99931 Database or file accessing
• CPC Class Codes: Y10S 707/99933 Query processing, i.e. searching
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation

Abstract

Methods and apparatus for household level segmentation are disclosed. An example method to classify consumers in clusters includes receiving population data indicative of a population of consumers and receiving a plurality of profiles, at least one profile to evaluate partitioning of the population of consumers. The example method also includes selecting at least one of the plurality of profiles based on a count limit value in accordance with a classification tree dimension split.

Claims

1. A segmentation system for classifying consumers in clusters, comprising:

a partitioning module, executed by a processor, to generate a plurality of classification trees, each of the plurality of classification trees including both behavioral and demographic consumer segmenting variables to classify a consumer population, each of the plurality of classification trees further including a plurality of decision nodes and a plurality of terminal nodes, each of the plurality of classification trees to produce a consumer cluster set having a plurality of consumer clusters represented by the terminal nodes, each decision node to indicate a portion of the consumer population and to split the portion of the consumer population into at least two other nodes in accordance with one of the consumer segmenting variables, the partitioning module to search the consumer cluster sets for an optimal consumer cluster set that optimizes a measure of the behavioral and demographic data, and consumers in each consumer cluster of the plurality of consumer clusters in the optimal consumer cluster set having similar behavioral and demographic characteristics to each other and at least one behavioral or demographic characteristic from consumers in other consumer clusters of the plurality of consumer clusters in the optimal consumer cluster set;

a profile definitions module, executed by the processor, to store profile definitions data to define evaluation profiles to evaluate partitioning of the consumer population, the partitioning module to determine a count for each of the decision nodes of each of the classification trees, at least one of the counts including a right split count, a left split count, and a total count for the corresponding decision node;

a profile data module, executed by the processor, to store summaries of the counts; and

a segment definitions module, executed by the processor, to store segment definitions data including variables used to define segments, the partitioning module to compare performance of the classification trees based on the stored profile definitions data, the summaries of the count, and the segment definitions data to determine the classification tree producing the optimal consumer cluster set, the consumer clusters in the optimal consumer cluster set are to focus marketing on groups of consumers.

2. A segmentation system as defined in claim 1, wherein the classification trees are to use Zhang'"'"'s methodology.

3. A segmentation system as defined in claim 1, wherein the partitioning module is to use a partitioning program.

4. A segmentation system as defined in claim 1, further comprising:

a summarization module to generate a summarization of data contained in the partitioning module; and

a summary data module to store the summarization of the data contained in the partitioning module.

5. A segmentation system as defined in claim 1, wherein the profile data module comprises a database to store profile definitions data.

6. A segmentation system as defined in claim 1, wherein the profile data module comprises an electronic file to store the profile data.

7. A segmentation system as defined in claim 1, wherein the segment definitions module comprises a database file to store segment definitions data.

8. A segmentation system as defined in claim 1, further comprising a cluster assignments module to store the plurality of consumer clusters generated by the partitioning module.

9. A segmentation system as defined in claim 8, wherein the cluster assignments module comprises a database table.

10. A method to classify consumers in clusters comprising:

receiving population data indicative of a population of consumers;

receiving a plurality of profiles, at least a first one of the profiles to evaluate partitioning of the population of consumers; and

selecting at least a second one of the plurality of profiles based on a count limit value in accordance with a classification tree dimension split to derive at least one of a node or a terminal node, each terminal node representing a partition of the population of consumers that is homogeneous with respect to both behavior and demographics.

11. A method as defined in claim 10, wherein selecting the at least the second one of the plurality of profiles based on the count limit value further comprises selecting the at least the second one of the plurality of profiles that exceeds the count limit value.

12. A method as defined in claim 10, further comprising removing from consideration at least a third one of the plurality of profiles that includes a count value lower than the count limit.

13. A method as defined in claim 10, further comprising retaining at least one of a plurality of the dimension splits based on a minimum population segment size.

14. A method as defined in claim 10, further comprising retaining at least one of a plurality of the dimension splits based on a split balance value.

15. A method as defined in claim 10, further comprising retaining at least one of a plurality of the dimension splits based on a Gini impurity measure.