Process and heuristic statistic for prospect selection through data mining

• US 7,725,415 B2
• Filed: 06/16/2008
• Issued: 05/25/2010
• Est. Priority Date: 12/19/2003
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Rollins, John Baxter, Hale, Curtis Richard
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): OLUDE AFOLABI, OLATOYOSI
• Application Number:
  US12/139,649
• Publication Number:
  US 20080249971A1
• Time in Patent Office: 708 Days
• Field of Search: 706/45, 706/46, 705/1, 705/10, 705/14, 705/26, 709/203
• US Class Current: 706/46
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0207 Discounts or incentives, e.g. coupons or rebates
• CPC Class Codes: G06Q 30/0244 Optimization
• CPC Class Codes: G06Q 30/0251 Targeted advertisements

Abstract

Method, system, and article of manufacture for selecting prospects for a product promotion though data mining. An initial set of prospects in a customer database is identified, by data mining, as initially identified prospects based on predetermined selection criteria. The number of initially identified prospects is compared to a target number of prospects. When the number of initially identified prospects matches the target number of prospects, the initially identified prospects are utilized as the final selection of prospects. When the number of initially identified prospects mismatches the target number of prospects, the final selection of prospects is determined by performing a culling process or an augmenting process to reduce or increase, respectively, the initial set of prospects using a heuristic measure H, until the number of prospects in the initial set of prospects matches the target number of prospects.

Claims

1. A method of identifying a final selection of prospects from a customer database containing a plurality of prospects for a product promotion, the method comprising:

using a data mining technique, identifying an initial set of prospects in a customer database as initially identified prospects; and

,

determining the final selection of prospects by performing one of a culling process and an augmenting process to, respectively, remove initially identified prospects from or add additional prospects of said customer database to the initial set of prospects using a heuristic measure H, until a count of prospects in said initial set of prospects matches a predetermined target number of prospects.

2. The method as set forth in claim 1, further including:

comparing the count of said initially identified prospects in said initial set of prospects to said target number of prospects;

when the count of initially identified prospects matches the target number of prospects, utilizing the initially identified prospects as the final selection of prospects; and

,

when the count of initially identified prospects mismatches the target number of prospects, determining the final selection of prospects by performing said one of the culling process and the augmenting process to, respectively, remove initially identified prospects from or add additional prospects of said customer database to the initial set of prospects using said heuristic measure H, until a count of prospects in said initial set of prospects matches a predetermined target number of prospects.

3. The method as set forth in claim 2, wherein:

said identifying includes;

using said data mining technique to identify at least one assigned cluster for each prospect of said initial set of prospects; and

,

using said data mining technique to identify at least one next-best cluster for each prospect of said initial set of prospects; and

,

said culling and augmenting processes include calculating said heuristic measure H by providing;

a confidence factor (Confidence) which is a measure of certainty that the prospect has been assigned to a best cluster;

a first score factor (Score1) which is a measure of how well the prospect fits in its assigned cluster; and

a second score factor (Score2) which is a measure of how well the prospect fits in a next-best cluster.

4. The method as set forth in claim 3, wherein:

the heuristic H=Confidence×

Score1÷

Score2.

5. The method as set forth in claim 4, wherein:

0≦

Confidence≦

1, and 0 is a lowest certainty;

0≦

Score1≦

1, and 0 represents a poorest best fit; and

0≦

Score2≦

1, and 0 represents a poorest next-best fit.

6. The method as set forth in claim 2, wherein said culling process includes:

ranking each prospect of said initial set of prospects by said heuristic measure H; and

selecting prospects from said initial set of prospects in order of H for the final selection of prospects until the target number is matched.

7. The method as set forth in claim 6, wherein the selecting prospects from the initial set in order of H includes selecting prospects from the initial set in descending order of H.

8. The method as set forth in claim 2, wherein the culling includes:

creating one or more sub-clusters of the initial set by utilizing said data mining technique;

ranking the sub-clusters by a respective potential for the promotion;

selecting from the sub-clusters in descending order of potential for the final selection of prospects until the target number is matched or would be exceeded by the last-selected sub-cluster; and

when the target number is not matched;

ranking each prospect of a particular sub-cluster by said heuristic measure H; and

selecting prospects from the particular sub-cluster in order of H for the final selection of prospects until the target number is matched.

9. The method as set forth in claim 8, wherein:

the selecting prospects from the particular sub-cluster in order of H includes selecting prospects from the particular sub-cluster in descending order of H.

10. The method as set forth in claim 8, wherein:

the ranking each prospect of a particular sub-cluster by said heuristic measure H includes ranking each prospect of the last-selected sub-cluster by said heuristic measure H.

11. The method as set forth in claim 2, wherein said augmenting process includes:

ranking prospects of the database not selected for inclusion in the initial set of prospects by said heuristic measure H as ranked prospects; and

selecting from the ranked prospects and adding the selected ranked prospects to the initial set in order of H for the final selection of prospects until the target number is matched.

12. The method as set forth in claim 11, wherein the selecting from the ranked prospects for addition to the initial set in order of H includes selecting from the ranked prospects for addition to the initial set in descending order of H.

13. The method as set forth in claim 2, wherein the augmenting process includes:

creating one or more sub-clusters of the remaining non-selected database prospects by data mining;

ranking the sub-clusters by a respective potential for the promotion;

selecting from the sub-clusters in descending order of potential for addition to the initial set as the final selection of prospects until the target number is matched or would be exceeded by the last-selected sub-cluster; and

when the target number is not matched;

ranking each prospect of a particular sub-cluster by a heuristic measure H; and

selecting prospects from the particular sub-cluster in order of H for addition to the initial set as the final selection of prospects until the target number is matched.

14. The method as set forth in claim 13, wherein:

the selecting prospects from the particular sub-cluster in order of H includes selecting prospects from the particular sub-cluster in descending order of H.

15. The method as set forth in claim 13, wherein:

the ranking each prospect of a particular sub-cluster by a heuristic measure H includes ranking each prospect of the last-selected sub-cluster by a heuristic measure H.

16. The method as set forth in claim 2, wherein:

said culling process includes;

ranking each prospect of the initial set by a heuristic measure H; and

selecting prospects from the initial set in order of H for the final selection of prospects until the target number is matched; and

,

said augmenting process includes;

ranking each remaining prospect of the database not selected for the initial set by a heuristic measure H; and

selecting prospects for addition to the initial set from the remaining prospects in descending order of H for the final selection of prospects until the target number is matched.

17. The method as set forth in claim 16 wherein the heuristic H comprises:

a confidence factor (Confidence) which is a measure of certainty that the prospect has been assigned to a best cluster;

a first score factor (Score

     1) which is a measure of how well the prospect fits in its assigned cluster; and

a second score factor (Score2) which is a measure of how well the prospect fits in a next-best cluster.

18. The method as set forth in claim 17, wherein:

the heuristic H=Confidence×

Score1÷

Score2.

19. The method as set forth in claim 18, wherein:

0≦

Confidence≦

1, and 0 is a lowest certainty;

0≦

Score1≦

1, and 0 represents a poorest best fit; and

0≦

Score2≦

1, and 0 represents a poorest next-best fit.

20. A system for identifying a final selection of prospects from a customer database containing a plurality of prospects for a product promotion, the system comprising:

data mining means for identifying an initial set of prospects in a customer database as initially identified prospects by utilizing a data mining technique; and

,

means for determining the final selection of prospects by performing one of a culling process means and an augmenting process means to, respectively, remove initially identified prospects from or add additional prospects of said customer database to the initial set of prospects using a heuristic measure H, until a count of prospects in said initial set of prospects matches a predetermined target number of prospects.

21. The system for identifying a final selection of prospects as set forth in claim 20, further including:

means for comparing the count of said initially identified prospects in said initial set of prospects to said target number of prospects;

means for utilizing the initially identified prospects as the final selection of prospects when the count of initially identified prospects matches the target number of prospects; and

,

means for determining the final selection of prospects by performing said one of the culling process means and the augmenting process means when the count of initially identified prospects mismatches the target number of prospects to, respectively, remove initially identified prospects from or add additional prospects of said customer database to the initial set of prospects using said heuristic measure H, until the count of prospects in said initial set of prospects matches said target number of prospects.

22. The system for identifying a final selection of prospects as set forth in claim 21, wherein:

said data mining means for identifying includes a data mining means to identify at least one assigned cluster for each prospect of said initial set of prospects and identify at least one next-best cluster for each prospect of said initial set of prospects; and

,

said culling process means and augmenting process means include means for calculating said heuristic measure H by providing;

a confidence factor (Confidence) which is a measure of certainty that the prospect has been assigned to a best cluster;

a first score factor (Score1) which is a measure of how well the prospect fits in its assigned cluster; and

a second score factor (Score2) which is a measure of how well the prospect fits in a next-best cluster.

23. The system for identifying a final selection of prospects as set forth in claim 22, wherein:

the heuristic H=Confidence×

Score1÷

Score2.

24. The system for identifying a final selection of prospects as set forth in claim 23, wherein:

0≦

Confidence≦

1, and 0 is a lowest certainty;

0≦

Score1≦

1, and 0 represents a poorest best fit; and

0≦

Score2≦

1, and 0 represents a poorest next-best fit.

25. The system for identifying a final selection of prospects as set forth in claim 21, wherein said culling process means includes:

means for ranking each prospect of the initial set of prospects by said heuristic measure H; and

means for selecting prospects from said initial set of prospects in order of H for the final selection of prospects until the target number is matched.

26. The system for identifying a final selection of prospects as set forth in claim 25, wherein the means for selecting prospects from the initial set in order of H includes means for selecting prospects from the initial set in descending order of H.

27. The system for identifying a final selection of prospects as set forth in claim 21, wherein the culling process means includes:

means for creating one or more sub-clusters of the initial set by utilizing said data mining technique;

means for ranking the sub-clusters by a respective potential for the promotion;

means for selecting from the sub-clusters in descending order of potential for the final selection of prospects until the target number is matched or would be exceeded by the last-selected sub-cluster; and

when the target number is not matched;

means for ranking each prospect of a particular sub-cluster by said heuristic measure H; and

means for selecting prospects from the particular sub-cluster in order of H for the final selection of prospects until the target number is matched.

28. The system for identifying a final selection of prospects as set forth in claim 27, wherein:

the means for selecting prospects from the particular sub-cluster in order of H includes means for selecting prospects from the particular sub-cluster in descending order of H.

29. The system for identifying a final selection of prospects as set forth in claim 27, wherein:

the means for ranking each prospect of a particular sub-cluster by said heuristic measure H includes means for ranking each prospect of the last-selected sub-cluster by said heuristic measure H.

30. The system for identifying a final selection of prospects as set forth in claim 21, wherein said augmenting process means includes:

means for ranking prospects of the database not selected for inclusion in the initial set of prospects by said heuristic measure H as ranked prospects; and

means for selecting from the ranked prospects and adding the selected ranked prospects to the initial set in order of H for the final selection of prospects until the target number is matched.

31. The system for identifying a final selection of prospects as set forth in claim 30, wherein the means for selecting prospects for addition to the initial set from the remaining prospects in order of H includes means for selecting prospects for addition to the initial set from the remaining prospects in descending order of H.

32. The system for identifying a final selection of prospects as set forth in claim 21, wherein the augmenting process means includes:

means for creating one or more sub-clusters of the remaining non-selected database prospects by data mining;

means for ranking the sub-clusters by a respective potential for the promotion;

means for selecting from the sub-clusters in descending order of potential for addition to the initial set as the final selection of prospects until the target number is matched or would be exceeded by the last-selected sub-cluster; and

when the target number is not matched;

means for ranking each prospect of a particular sub-cluster by a heuristic measure H; and

means for selecting prospects from the particular sub-cluster in order of H for addition to the initial set as the final selection of prospects until the target number is matched.

33. The system for identifying a final selection of prospects as set forth in claim 32, wherein:

the means for selecting prospects from the particular sub-cluster in order of H includes means for selecting prospects from the particular sub-cluster in descending order of H.

34. The system for identifying a final selection of prospects as set forth in claim 32, wherein:

the means for ranking each prospect of a particular sub-cluster by a heuristic measure H includes means for ranking each prospect of the last-selected sub-cluster by a heuristic measure H.

35. The system for identifying a final selection of prospects as set forth in claim 21, wherein:

said culling process means includes;

means for ranking each prospect of the initial set by a heuristic measure H; and

means for selecting prospects from the initial set in order of H for the final selection of prospects until the target number is matched; and

,

said augmenting process means includes;

means for ranking each remaining prospect of the database not selected for the initial set by a heuristic measure H; and

means for selecting prospects for addition to the initial set from the remaining prospects in descending order of H for the final selection of prospects until the target number is matched.

36. The system for identifying a final selection of prospects as set forth in claim 35 wherein the heuristic H comprises:

a confidence factor (Confidence) which is a measure of certainty that the prospect has been assigned to a best cluster;

a first score factor (Score1) which is a measure of how well the prospect fits in its assigned cluster; and

a second score factor (Score2) which is a measure of how well the prospect fits in a next-best cluster.

37. The system for identifying a final selection of prospects as set forth in claim 36, wherein:

the heuristic H=Confidence×

Score1÷

Score2.

38. The system for identifying a final selection of prospects as set forth in claim 37, wherein:

0≦

Confidence≦

1, and 0 is a lowest certainty;

0≦

Score1≦

1, and 0 represents a poorest best fit; and

0≦

Score2≦

1, and 0 represents a poorest next-best fit.

39. A computer program product comprising a computer usable medium having computer readable code embodied in said medium for use in identifying a final selection of prospects from a customer database containing a plurality of prospects for a product promotion, comprising:

data mining program code that identifies an initial set of prospects in a customer database as initially identified prospects by utilizing a data mining technique; and

,

program code that determines the final selection of prospects by utilizing one of a culling process program code and an augmenting process program code to, respectively, remove initially identified prospects from or add additional prospects of said customer database to the initial set of prospects using a heuristic measure H, until a count of prospects in said initial set of prospects matches a predetermined target number of prospects.

40. The computer program product as set forth in claim 39, further including:

program code that compares the count of said initially identified prospects in said initial set of prospects to said target number of prospects;

program code that utilizes the initially identified prospects as the final selection of prospects when the count of initially identified prospects matches the target number of prospects; and

,

program code that determines the final selection of prospects by utilizing said one of the culling process program code and the augmenting process program code when the count of initially identified prospects mismatches the target number of prospects to, respectively, remove initially identified prospects from or add additional prospects of said customer database to the initial set of prospects using said heuristic measure H, until the count of prospects in said initial set of prospects matches said target number of prospects.

41. The computer program product as set forth in claim 40, wherein:

said data mining program code that identifies includes data mining program code that identifies at least one assigned cluster for each prospect of said initial set of prospects and identifies at least one next-best cluster for each prospect of said initial set of prospects; and

,

said program code that determines the final selection of prospects includes program code for calculating said heuristic measure H by providing;

a confidence factor (Confidence) which is a measure of certainty that the prospect has been assigned to a best cluster;

a first score factor (Score1) which is a measure of how well the prospect fits in its assigned cluster; and

a second score factor (Score2) which is a measure of how well the prospect fits in a next-best cluster.

42. The computer program product as set forth in claim 41, wherein:

the heuristic H=Confidence×

Score1÷

Score2.

43. The computer program product as set forth in claim 42, wherein:

0≦

Confidence≦

1, and 0 is a lowest certainty;

0≦

Score1≦

1, and 0 represents a poorest best fit; and

0≦

Score2≦

1, and 0 represents a poorest next-best fit.

44. The computer program product as set forth in claim 40, wherein said culling process program code includes:

program code that ranks each prospect of said initial set of prospects by said heuristic measure H; and

program code that selects prospects from said initial set of prospects in order of H for the final selection of prospects until the target number is matched.

45. The computer program product as set forth in claim 44, wherein the program code that selects prospects from the initial set in order of H includes program code that selects prospects from the initial set in descending order of H.

46. The computer program product as set forth in claim 40, wherein said culling process program code includes:

program code that creates one or more sub-clusters of the initial set by utilizing said data mining technique;

program code that ranks the sub-clusters by a respective potential for the promotion;

program code that selects from the sub-clusters in descending order of potential for the final selection of prospects until the target number is matched or would be exceeded by the last-selected sub-cluster; and

when the target number is not matched;

program code that ranks each prospect of a particular sub-cluster by said heuristic measure H; and

program code that selects prospects from the particular sub-cluster in order of H for the final selection of prospects until the target number is matched.

47. The computer program product as set forth in claim 46, wherein:

the program code that selects prospects from the particular sub-cluster in order of H includes program code that selects prospects from the particular sub-cluster in descending order of H.

48. The computer program product as set forth in claim 46, wherein:

the program code that ranks each prospect of a particular sub-cluster by said heuristic measure H includes program code that ranks each prospect of the last-selected sub-cluster by said heuristic measure H.

49. The computer program product as set forth in claim 40, wherein said augmenting process program code includes:

program code that ranks prospects of the database not selected for inclusion in the initial set of prospects by said heuristic measure H as ranked prospects; and

program code that selects from the ranked prospects and adds the selected ranked prospects to the initial set in order of H for the final selection of prospects until the target number is matched.

50. The computer program product as set forth in claim 49, wherein the program code that selects prospects for addition to the initial set from the remaining prospects in order of H includes program code that selects prospects for addition to the initial set from the remaining prospects in descending order of H.

51. The computer program product as set forth in claim 40, wherein said augmenting process program code includes:

program code that creates one or more sub-clusters of the remaining non-selected database prospects by data mining;

program code that ranks the sub-clusters by a respective potential for the promotion;

program code that selects from the sub-clusters in descending order of potential for addition to the initial set as the final selection of prospects until the target number is matched or would be exceeded by the last-selected sub-cluster; and

when the target number is not matched;

program code that ranks each prospect of a particular sub-cluster by a heuristic measure H; and

program code that selects prospects from the particular sub-cluster in order of H for addition to the initial set as the final selection of prospects until the target number is matched.

52. The computer program product as set forth in claim 51, wherein:

the program code that selects prospects from the particular sub-cluster in order of H includes program code that selects prospects from the particular sub-cluster in descending order of H.

53. The computer program product as set forth in claim 51, wherein:

the program code that ranks each prospect of a particular sub-cluster by a heuristic measure H includes program code that ranks each prospect of the last-selected sub-cluster by a heuristic measure H.

54. The computer program product as set forth in claim 40, wherein:

said culling process program code includes;

program code that ranks each prospect of the initial set by a heuristic measure H; and

program code that selects prospects from the initial set in order of H for the final selection of prospects until the target number is matched; and

,

said augmenting process program code includes;

program code that ranks each remaining prospect of the database not selected for the initial set by a heuristic measure H; and

program code that selects prospects for addition to the initial set from the remaining prospects in descending order of H for the final selection of prospects until the target number is matched.

55. The computer program product as set forth in claim 54 wherein the heuristic H comprises:

a confidence factor (Confidence) which is a measure of certainty that the prospect has been assigned to a best cluster;

a first score factor (Score1) which is a measure of how well the prospect fits in its assigned cluster; and

a second score factor (Score2) which is a measure of how well the prospect fits in a next-best cluster.

56. The computer program product as set forth in claim 55, wherein:

the heuristic H=Confidence×

Score1÷

Score2.

57. The computer program product as set forth in claim 56, wherein:

0≦

Confidence≦

1, and 0 is a lowest certainty;

0≦

Score1≦

1, and 0 represents a poorest best fit; and

0≦

Score2≦

1, and 0 represents a poorest next-best fit.