Targeted incentives based upon predicted behavior
View Patent ↗A system and method for anticipating consumer behavior and determining transaction incentives for influencing consumer behavior comprises a computer system and associated database for determining cross time correlations between transaction behavior, for applying the function derived from the correlations to consumer records to predict future consumer behavior, and for deciding on transaction incentives to offer the consumers based upon their predicted behavior.
1. A computer system, comprising:
a digital processor, memory, a device for inputting information from a user, and an output for outputting information;
wherein said computer system stores in said memory, data defining:
(1) values representing statistical correlations;
(2) a first plurality of consumer records, each including transaction data indicating identification of product purchased and date of purchase;
(3) a predictive model function; and
(4) a second consumer record for a second consumer including transaction data and a first correlated class predictive data field;
wherein said computer system is programmed to determine said values representing statistical correlations from said first plurality of consumer records by, first, determining values representing correlations of data within each one of said first plurality of consumer records between (a) and (b), wherein: (a) are transactions in at least one transaction class for transactions that occurred during at least one first time period and (b) are transactions in at least one transaction class that occurred during at least one second time period, said at least one second time period being subsequent in time to said at least one first time period, and then, second, determining said values representing statistical correlations from said values representing correlations;
wherein said predictive model function is defined at least in part by said values representing statistical correlations;
wherein said computer system is programmed to apply said predictive model function to transaction data of said second consumer record for transactions that occurred during a third time period, to result in first correlated class predictive data;
wherein said computer system is programmed to store said first correlated class predictive data in said first correlated class predictive data field of said second consumer record.
2. The system of claim 1 wherein said determining values representing correlations does not include determining cross-correlations.
3. The system of claim 1 wherein said determining values representing correlations does include determining cross-correlations.
4. The system of claim 1 wherein said computer system is programmed to determine at least one value representing correlations between different transaction classes.
5. The system of claim 1 wherein said computer system is programmed to determine at least one value representing correlations between all data from a genus of transaction classes and all data associated with one specie of that genus.
6. The system of claim 1 programmed to determine whether to offer a transaction incentive to said second consumer based upon data in said first correlated class predictive data field of said second consumer record.
7. A computer implemented method for using a computer system comprising a digital processor, memory, a device for inputting information from a user, and an output for outputting information, comprising:
storing in said memory data defining:
(1) values representing statistical correlations;
(2) a first plurality of consumer records, each including transaction data indicating identification of product purchased and date of purchase;
(3) a predictive model function; and
(4) a second consumer record for a second consumer including transaction data and a first correlated class predictive data field;
determining, using said computer system, values representing statistical correlations from said first plurality of consumer records by, first, determining values representing correlations of data within each one of said first plurality of consumer records between (a) and (b), wherein: (a) are transactions in at least one transaction class for transactions that occurred during at least one first time period and (b) are transactions in at least one transaction class that occurred during at least one second time period, said at least one second time period being subsequent in time to said at least one first time period, and then, second, determining said values representing statistical correlations from said values representing correlations;
wherein said predictive model function is defined at least in part by said values representing statistical correlations;
applying, using said computer system, said predictive model function to transaction data of said second consumer record for transactions that occurred during a third time period, to result in first correlated class predictive data;
storing, in said memory of said computer system, said first correlated class predictive data in said first correlated class predictive data field of said second consumer record.
8. A non transitory computer readable medium containing a computer program product for predictive modeling, the computer program product comprising:
program code for storing in memory of a computer system, data defining:
(1) values representing statistical correlations;
(2) a first plurality of consumer records, each including transaction data indicating identification of product purchased and date of purchase;
(3) a predictive model function; and
(4) a second consumer record for a second consumer including transaction data and a first correlated class predictive data field;
program code for determining, using said computer system, values representing statistical correlations from said first plurality of consumer records by, first, determining values representing correlations of data within each one of said first plurality of consumer records between (a) and (b), wherein: (a) are transactions in at least one transaction class for transactions that occurred during at least one first time period and (b) are transactions in at least one transaction class that occurred during at least one second time period, said at least one second time period being subsequent in time to said at least one first time period, and then, second, determining said values representing statistical correlations from said values representing correlations;
wherein said predictive model function is defined at least in part by said values representing statistical correlations;
program code for applying said predictive model function to transaction data of said second consumer record for transactions that occurred during a third time period, to result in first correlated class predictive data;
program code for storing in said memory said first correlated class predictive data in said first correlated class predictive data field of said second consumer record.