System and method for predicting card member spending using collaborative filtering

• US 8,204,784 B2
• Filed: 06/22/2010
• Issued: 06/19/2012
• Est. Priority Date: 12/28/2004
• Status: Expired due to Fees

• Current Assignee: Liberty Peak Ventures, LLC (Dominion Harbor Enterprises, LLC)
• Original Assignee: American Express Travel Related Services Company, Inc. (American Express Company)
• Inventors: Karmali, Siddhesh, Baqri, Ali H., Marshall, Vernon, Garrett, Rachel B., Bhagchandani, Puneet B.
• Primary Examiner(s): SORKOWITZ, DANIEL M
• Application Number:
  US12/820,713
• Publication Number:
  US 20100256982A1
• Time in Patent Office: 728 Days
• Field of Search: 705/14.52, 705/14.53, 705/14.49, 705/14.4
• US Class Current: 705/14.52
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/0251 Targeted advertisements
• CPC Class Codes: G06Q 30/0254 based on statistics
• CPC Class Codes: G06Q 30/0255 based on user history
• CPC Class Codes: G06Q 30/0261 based on user location
• CPC Class Codes: G06Q 30/04 Billing or invoicing
• CPC Class Codes: G06Q 50/12 Hotels or restaurants

Abstract

The disclosed method and system allows a credit or charge card issuer to provide its card members with a list of restaurants that might be of interest based on the financial transactions of similar card members. In one instance, this method filters financial transaction data from a plurality of card members that involves a plurality of restaurants to generate a set of candidate restaurant recommendations for a selected card member. This set of candidate restaurant recommendations is processed to yield a list of restaurant recommendations for the selected customer that is prioritized on the basis of the selected card member accepting the recommendation. The list of restaurant recommendations is then reported to the selected card member to enhance card use and marketing.

Claims

1. A method, comprising:

filtering, by a computer for filtering financial transaction data, a data set to generate a set of candidate merchants for the selected account holder, wherein the data set corresponds to a selected account holder of a plurality of account holders, and wherein the data set corresponds to financial transactions of the plurality of account holders with a plurality of merchants; and

generating, by the computer, for the selected account holder, a probability of acceptance for each of the candidate merchants, wherein the probability of acceptance (P) is determined by;

P=A+(B*A−

K)), wherein;

A is a number of account holders having transactions with both the candidate merchant and an identified merchant;

K is a constant;

B is a value corresponding to a ratio of A to a total number of account holder transactions with the identified merchant; and

a value of (A−

K) is set to zero if its determined value results in a negative number.

2. The method of claim 1, Wherein the filtering further comprises:

identifying merchants within the plurality of merchants that have at least one financial transaction with the selected account holder over a period of time;

determining, for each non-identified merchant within the plurality of merchants, a strength of association between each identified merchant and each non-identified merchant;

generating the set of candidate merchants based on at least the strengths of association; and

filtering the set of candidate merchants according to at least one of;

(i) an average transaction size and (ii) geographic location.

3. The method of claim 2, wherein the filtering the set of candidate merchants according to average transaction size further comprises:

obtaining an average transaction size for each candidate merchant over the period of time;

determining an average transaction size for the identified merchants over the period of time; and

eliminating merchants from the set of candidate merchants based on at least the average transaction size of the candidate merchant.

4. The method of claim 3, further comprising before the generating of a probability of acceptance, eliminating merchants from the set of candidate merchants whose average transaction size varies from the computed average transaction size by more than a predetermined amount.

5. The method of claim 2, wherein the filtering the set of candidate merchants according to location further comprises:

obtaining, for each identified merchant, a distance traveled by the selected account holder when visiting the identified merchant;

determining a standard deviation for the obtained travel distances; and

eliminating merchants from the set of candidate merchants based on at least the computed standard deviation of the obtained travel distances.

6. The method of claim 5, further comprising, before the generating of a probability of acceptance, eliminating merchants from the set of candidate merchants based on the distance between the candidate merchant and the account holder billing address exceeding the computed average distance by a predetermined amount.

7. The method of claim 1, wherein the generating further comprises prioritizing each merchant within the set of candidate merchants based on the computed probability of acceptance of the candidate merchant to generate the list of recommended merchants.

8. The method of claim 1, further comprising:

determining, for each identified merchant, a probability that the selected account holder will have future financial transactions with the identified merchant;

determining, for each identified merchant and for each candidate merchant, a probability that account holders within the plurality of account holders will have financial transactions at both the identified and the candidate merchants; and

determining, for each candidate merchant, the probability that the selected account holder will have multiple financial transactions at the candidate merchant based on;

(i) the probability that the selected account holder will have future financial transactions with the identified merchant and (ii) the probability that the other account holders will have financial transactions at both the identified and the candidate merchant,

wherein the probability of acceptance represents the probability that the selected account holder will have multiple future financial transactions involving the candidate merchant.

9. The method of claim 8, wherein the probability that the selected account holder will have a future financial transaction with the identified merchant represents a ratio of a total number of financial transactions involving the selected account holder and the identified merchant to a total number of financial transactions involving the selected account holder.

10. The method of claim 8, wherein the probability that account holders will have financial transactions at both the identified merchant and the candidate merchant represents a ratio of a number of financial transactions involving the identified merchant and the candidate merchant to the total number of financial transactions involving the identified merchant,

11. The method of claim 1, further comprising determining an adventurousness measure of the account holder at least partially based on the determined probability of acceptance for the candidate merchant.

12. A computer readable storage medium comprising a non-transitory, tangible computer useable storage medium having computer executable instructions recorded thereon, when executed by a computer for filtering financial transaction data, cause the computer to perform operations, comprising:

filtering, by the computer, a data set to generate a set of candidate merchants for the selected account holder, wherein the data set corresponds to a selected account holder of a plurality of account holders, and wherein the data set corresponds to financial transactions of the plurality of account holders with a plurality of merchants; and

generating, by the computer, for the selected account holder, a probability of acceptance for each of the candidate merchants, wherein the probability of acceptance (P) is determined by;

P=A+(B*(A−

K)), wherein;

A is a number of account holders having transactions with both the candidate merchant and an identified merchant;

K is a constant;

B is a value corresponding to a ratio of A to a total number of account holder transactions with the identified merchant; and

a value of (A−

K) is set to zero if its determined value results in a negative number.

13. The medium of claim 12, wherein the filtering further comprises:

identifying merchants within the plurality of merchants that have at least one financial transaction with the selected account holder over a period of time;

determining, for each non-identified merchant within the plurality of merchants, a strength of association between each identified merchant and each non-identified merchant;

generating the set of candidate merchants based on at least the strengths of association; and

filtering the set of candidate merchants according to at least one of;

(i) an average transaction size and (ii) geographic location.

14. The medium of claim 13, wherein the filtering the set of candidate merchants according to average transaction size further comprises:

obtaining an average transaction size for each candidate merchant over the period of time;

determining an average transaction size for the identified merchants over the period of time; and

eliminating merchants from the set of candidate merchants based on at least the average transaction size of the candidate merchant.

15. The medium of claim 14, further comprising before the generating of a probability of acceptance, eliminating merchants from the set of candidate merchants whose average transaction size varies from the computed average transaction size by more than a predetermined amount.

16. The medium of claim 13, wherein the filtering the set of candidate merchants according to location further comprises:

obtaining, for each identified merchant, a distance traveled by the selected account holder when visiting the identified merchant;

determining a standard deviation for the obtained travel distances; and

eliminating merchants from the set of candidate merchants based on at least the computed standard deviation of the obtained travel distances.

17. The medium of claim 16, further comprising, before the generating of a probability of acceptance, eliminating merchants from the set of candidate merchants based on the distance between the candidate merchant and the account holder billing address exceeding the computed average distance by a predetermined amount.

18. The medium of claim 12, wherein the generating further comprises prioritizing each merchant within the set of candidate merchants based on the computed probability of acceptance of the candidate merchant to generate the list of recommended merchants.

19. The medium of claim 12, further comprising:

determining, for each identified merchant, a probability that the selected account holder will have future financial transactions with the identified merchant;

determining, for each identified merchant and for each candidate merchant, a probability that account holders within the plurality of account holders will have financial transactions at both the identified and the candidate merchants; and

determining, for each candidate merchant, the probability that the selected account holder will have multiple financial transactions at the candidate merchant based on;

(i) the probability that the selected account holder will have future financial transactions with the identified merchant and (ii) the probability that the other account holders will have financial transactions at both the identified and the candidate merchant,

wherein the probability of acceptance represents the probability that the selected account holder will have multiple future financial transactions involving the candidate merchant.

20. A computer-based system for predicting spending habits of account holders, comprising:

a processor for filtering financial transaction data; and

an instruction memory in communication with the processor for storing a plurality of processing instructions for directing the processor to;

filter, by the processor, a data set to generate a set of candidate merchants for the selected account holder, wherein the data set corresponds to as selected account holder of a plurality of account holders, and wherein the data set corresponds to financial transactions of the plurality of account holders with a plurality of merchants; and

generate, by the processor and for the selected account holder, a probability of acceptance for each of the candidate merchants, wherein the probability of acceptance (P) is determined by;

P=A+(B*(A−

K)), wherein;

A is a number of account holders having transactions with both the candidate merchant and an identified merchant;

K is a constant;

B is a value corresponding to as ratio of A to a total number of account holder transactions with the identified merchant; and

a value of (A−

K) is set to zero if its determined value results in a negative number.