Computer-Implemented Data Storage Systems And Methods For Use With Predictive Model Systems

• US 20090192957A1
• Filed: 04/03/2009
• Published: 07/30/2009
• Est. Priority Date: 03/24/2006
• Status: Abandoned Application

• Current Assignee: Chao-Wen Chen, Paul C. Dulany, Radu Drossu, Revathi Subramanian
• Original Assignee: Chao-Wen Chen, Paul C. Dulany, Radu Drossu, Revathi Subramanian
• Inventors: Dulany, Paul C., Drossu, Radu, Subramanian, Revathi, Chen, Chao-Wen (Kevin)
• Application Number:
  US12/418,174
• Publication Number:
  US 20090192957A1
• Time in Patent Office: Days
• US Class Current: 706/21
• 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 20/4016 involving fraud or risk level assessment in transaction processing
• CPC Class Codes: G06Q 30/0185 Product, service or business identity fraud
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities
• CPC Class Codes: G06Q 30/06 Buying, selling or leasing transactions
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof
• CPC Class Codes: G06Q 40/06 Asset management; Financial planning or analysis
• CPC Class Codes: G06Q 40/12 Accounting

Abstract

Systems and methods for performing fraud detection. As an example, a system and method can be configured to contain a raw data repository for storing raw data related to financial transactions. A data store contains rules to indicate how many generations or to indicate a time period within which data items are to be stored in the raw data repository. Data items stored in the raw data repository are then accessed by a predictive model in order to perform fraud detection.

Claims

1. A computer-implemented method for imputing missing values for use in a fraud analysis predictive model, said method comprising:

executing instructions on a computer system for receiving an input data set from a computer-readable data store wherein one or more values are missing with respect to one or more attributes of the input data set;

executing instructions on the computer system for determining a value for an attribute based upon a closed form equation representing the value to be used when encountering a missing value for the attribute;

wherein the closed form equation is based upon fraud-related tag information;

executing instructions on the computer system for using in the fraud analysis predictive model the determined value;

wherein the receiving an input data set, determining a value for an attribute, and using the determined value are performed on a data processor.

2. The method of claim 1, wherein the fraud-related tag considerations include a tag associated with an input record as to whether the input record is considered fraud or non-fraud.

3. The method of claim 1, wherein the fraud analysis predictive model is a neural network predictive model.

4. The method of claim 1, wherein the closed form equation is generated for an attribute based upon historical data associated with the attribute.

5. The method of claim 4, wherein the closed form equation is generated by using an optimality criterion involving the tag information associated with the attributes.

6. The method of claim 1, wherein the input data set includes fraud-related features;

wherein the closed form equation is independent of the features.

7. The method of claim 1, wherein the missing data values pertain to missing authorization data values, missing posting data values, missing settlement data values, missing statement data values, missing payment data values, or missing non-monetary data values.

8. The method of claim 1, wherein the determined missing value is used in training, testing, or training and testing the fraud analysis predictive model.

9. The method of claim 8, wherein the training of the fraud analysis predictive model includes:

receiving training data;

training a first predictive model using the received training data;

said training data including data from a plurality of accounts;

using a partitioning criterion to determine how to partition the training data into partitions;

said partitioning criterion being based upon a fraud ranking violation metric;

training multiple predictive models sequentially using at least one of the partitions of training data;

said training of predictive models being performed to optimize a fraud ranking violation;

wherein the first and second predictive models are combined for use in predicting credit card or debit card fraud;

wherein said receiving training data, training a first predictive model, determining how to partition the training data, and training multiple predictive models are performed on a data processor.

10. The method of claim 9, wherein the first and second predictive models are neural network predictive models.

11. The method of claim 10, wherein the determined missing value is used as part of the input layer to the combined first and second predictive models.

12. The method of claim 1, wherein the determined missing value is used in a lookup table by the fraud analysis predictive model to predict in a production environment whether an account of an entity has been compromised.

13. A computer-implemented system for imputing missing values for use in a fraud analysis predictive model, comprising:

a computer-readable data store to store an input data set wherein one or more values are missing with respect to one or more attributes of the input data set;

value determination software instructions on a computer-readable medium to determine a value for an attribute based upon a closed form equation representing the value to be used when encountering a missing value for the attribute;

wherein the closed form equation is based upon fraud-related tag information;

wherein the determined value is for use in the fraud analysis predictive model.