System and method for performing risk and credit analysis of financial service applications

• US 5,696,907 A
• Filed: 02/27/1995
• Issued: 12/09/1997
• Est. Priority Date: 02/27/1995
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Tom, Mohimm Daniel
• Primary Examiner(s): Teska, Kevin J.
• Assistant Examiner(s): WALDER, STEPHEN JOSEPH
• Application Number:
  US08/395,213
• Time in Patent Office: 1,016 Days
• Field of Search: 364/408, 395/20-23, 395/51, 395/61, 395/75, 395/76, 395/238
• US Class Current: 705/38
• CPC Class Codes: G06Q 40/03 Credit; Loans; Processing thereof
• CPC Class Codes: G06Q 40/08 Insurance

Abstract

The present invention discloses a method and system for performing risk and credit analysis of financial service applications with a neural network. The neural network imitates and perfects a credit manager'"'"'s evaluation and decision process to control loss and guide business expansion. In particular, the neural network screens applications to control loss and to find directions where business volume can increase with a minimum increase in loss. Initially, data variables are pre-processed and applied to the neural network. The neural network in the present invention is optimized by a non-iterative regression process, as opposed to the computationally intensive back propagation algorithm.

Claims

1. A computer implemented method for performing risk and credit analysis of financial service applications with a neural network having an input layer of processing nodes, an output layer having a processing node, and an intermediate layer of processing nodes coupling the input layer to the output layer by weighted connections, the method comprising the steps of:

collecting data and status information from a plurality of previously approved financial service applications;

applying the data into the input layer of the neural network;

organizing the data at the input layer of the neural network into a plurality of groups, each of the plurality of groups containing variables used to perform risk and credit analysis;

classifying the grouped data from each of the plurality of groups into ordinal values and categorical values;

applying each of the plurality of groups of data from the processing nodes of the input layer to a separate processing node of the intermediate layer of the neural network;

applying the data from each of the separate processing nodes at the intermediate layer of the neural network to the processing node of .the output layer;

optimizing the weighted connections of the neural network to an approval criteria that increases approval volume of financial service applications with a minimum loss; and

providing data from a recently filed financial service application to the optimized neural network for evaluation.

2. The method according to claim 1, wherein the status information from the plurality of previously approved financial service applications comprises a termination due to default in payment, an early termination by a financial service applicant, a normal termination, and a financial service account open without any delinquencies in payment.

3. The method according to claim 2, further comprising the step of arranging the status information into an acceptable class and an unacceptable class, the acceptable class comprising a normal termination, and a financial service account open without any delinquencies in payment, the unacceptable class comprising a termination due to default in payment and a early termination by a financial service applicant.

4. The method according to claim 3, further comprising the step of determining vulnerability for the data in each of the plurality of groups at the input layer.

5. The method according to claim 1, wherein the plurality of groups applied to the separate processing nodes in the intermediate layer comprise employment stability, residential stability, credit history, debt and income history, and miscellaneous variables.

6. The method according to claim 1, further comprising the step of sorting the ordinal values for each of the plurality of groups in an ordered ranking from a highest ordinal value to a lowest ordinal value, the highest ordinal value assigned a one, the lowest ordinal value assigned a zero, and values in between the highest and lowest ordinal value assigned a value scaled linearly between zero and one.

7. The method according to claim 1, wherein the weighted connections are determined according to a multiple regression analysis using the data and status information, the weighted connections are sorted, ranked, and normalized to values between zero and one.

8. The method according to claim 7, further comprising sending the weighted connections to a credit manager for examination and for inferring a direction of relaxation to achieve business expansion while controlling loss.

9. The method according to claim 1, further comprising the step of screening the recently filed financial service application in accordance with the approval criteria, the recently filed financial service application approved if risk is minimal and rejected if risk is high.

10. A system for performing risk and credit analysis of financial service applications comprising:

a data storage for storing data and status information from a plurality of previously approved financial service applications;

means for pre-processing the data into a plurality of groups, each of the plurality of groups containing variables used to perform risk and credit analysis;

a neural network having an input layer of processing nodes, an output layer having a processing node, and an intermediate layer of processing nodes coupling the input layer to the output layer by weighted connections, the pre-processed data received at the input layer of the neural network, the input layer of the neural network organizing the data into a plurality of groups, each of the plurality of groups containing variables used to perform risk and credit analysis;

each of the plurality of groups of data from the processing nodes of the input layer applied to a separate processing node of the intermediate layer of, the data from each of the separate processing nodes at the intermediate layer applied to the processing node of the output layer of the neural network;

means for optimizing the neural network to an approval criteria that increase approval volume of financial service applications; and

a processor coupled to the optimized neural network for providing data from a recently filed financial service application to the optimized neural network, the optimized neural network determining the risk of the recently filed financial service application in accordance with the approval criteria.

11. The system according to claim 10, wherein the pre-processing means classifies the grouped data from each of the plurality of groups into ordinal values and categorical values.

12. The system according to claim 11, wherein the status information from the plurality of previously approved financial service applications comprises a termination due to default in payment, an early termination by a financial service applicant, a normal termination, and a financial service account open without any delinquencies in payment.

13. The system according to claim 12, further comprising means for arranging the status information into an acceptable class and an unacceptable class, the acceptable class comprising a normal termination, and a financial service account open without any delinquencies in payment, the unacceptable class comprising termination due to default in payment and an early termination by a financial service applicant.

14. The system according to claim 13, further comprising means for determining vulnerability for the data in each of the plurality of groups at the input layer.

15. The system according to claim 11, further comprising means for sorting the ordinal values for each of the plurality of groups in an ordered ranking from a highest ordinal value to a lowest ordinal value, the highest ordinal value assigned a one, the lowest ordinal value assigned a zero, and values in between the highest and lowest ordinal value assigned a value scaled linearly between zero and one.

16. The system according to claim 10, wherein the plurality of groups applied to the separate processing nodes in the intermediate layer comprise employment stability, residential stability, credit history, debt and income history, and miscellaneous variables.

17. The system according to claim 10, wherein the weighted connections are determined according to a multiple regression analysis using the data and status information, the weighted connections are sorted, ranked, and normalized to values between zero and one.

18. The system according to claim 17, further comprising means for sending the weighted connections to a credit manager for examination and for inferring a direction of relaxation to achieve business expansion while controlling loss.

19. A computer implemented method for performing risk and credit analysis of financial service applications comprising the steps of:

collecting data from a plurality of previously approved financial service applications;

organizing the data into a plurality of groups each containing variables used to perform risk and credit analysis;

constructing a neural network having an input layer of processing nodes, an output layer having a processing node, and an intermediate layer of processing nodes coupling the input layer to the output layer, wherein the input layer of processing nodes are separated according to each of the plurality of groups and each of the groups are applied to a separate processing node in the intermediate layer;

determining weighted connections between the processing nodes in the input layer, the intermediate layer, and the output layer of the neural network with a non-iterative regression;

optimizing the weighted connections of the neural network for inferring a direction to achieve business expansion; and

providing data from a recently filed financial service application to the optimized neural network for evaluation.

20. The computer-implemented method according to claim 19, further comprising the step of classifying the grouped data from each of the plurality of groups into ordinal values and categorical values.

21. The computer-implemented method according to claim 20, further comprising the step of sorting the ordinal values for each of the plurality of groups in an ordered ranking from a highest ordinal value to a lowest ordinal value, the highest ordinal value assigned a one, the lowest ordinal value assigned a zero, and values in between the highest and lowest ordinal value assigned a value scaled linearly between zero and one.

22. The computer-implemented method according to claim 19, further comprising the step of determining status information from the plurality of previously approved financial service applications, wherein the status information comprises a termination due to default in payment, an early termination by a financial service applicant, a normal termination, and a financial service account open without any delinquencies in payment.

23. The computer-implemented method according to claim 22, further comprising the step of arranging the status information into an acceptable class and an unacceptable class, the acceptable class comprising a normal termination, and a financial service account open without any delinquencies in payment, the unacceptable class comprising a termination due to default in payment and an early termination by a financial service applicant.

24. The computer-implemented method according to claim 23, further comprising the step of determining vulnerability for the data in each of the plurality of groups at the input layer.

25. The computer-implemented method according to claim 19, wherein the plurality of groups applied to the separate processing nodes in the intermediate layer comprise employment stability, residential stability, credit history, debt and income history, and miscellaneous variables.

26. The computer-implemented method according to claim 19, further comprising the step of sorting, ranking, and normalizing the weighted connections to values between zero and one.

27. A system for performing risk and credit analysis of financial service applications comprising:

a data storage for storing data from a plurality of previously approved financial service applications;

means for pre-processing the data into a plurality of groups, each of the plurality of groups containing variables used to perform risk and credit analysis;

means for constructing a neural network having an input layer of processing nodes, an output layer having a processing node, and an intermediate layer of processing nodes coupling the input layer to the output layer, wherein the input layer of processing nodes are separated according to each of the plurality of groups and each of the groups are applied to a separate processing node in the intermediate layer;

means for determining weighted connections between the processing nodes in the input layer, the intermediate layer, and the output layer of the neural network with a non-iterative regression;

means for optimizing the weighted connections of the neural network for inferring a direction to achieve business expansion; and

a processor coupled to the optimized neural network for providing data from a recently filed financial service application to the optimized neural network, the optimized neural network determining the risk of the recently filed financial service application.

28. The system according to claim 27, wherein the pre-processing means classifies the grouped data from each of the plurality of groups into ordinal values and categorical values.

29. The system according to claim 28, wherein the pre-processing means sorts the ordinal values for each of the plurality of groups in an ordered ranking from a highest ordinal value to a lowest ordinal value, the highest ordinal value assigned a one, the lowest ordinal value assigned a zero, and values in between the highest and lowest ordinal value assigned a value scaled linearly between zero and one.

30. The system according to claim 27, wherein the pre-processing means determines status information from the plurality of previously approved financial service applications, wherein the status information comprises a termination due to default in payment, an early termination by a financial service applicant, a normal termination, and a financial service account open without any delinquencies in payment.

31. The system according to claim 30, wherein the pre-processing means arranges the status information into an acceptable class and an unacceptable class, the acceptable class comprising a normal termination, and a financial service account open without any delinquencies in payment, the unacceptable class comprising a termination due to default in payment and a early termination by a financial service applicant.

32. The system according to claim 31, wherein the pre-processing means determines vulnerability for the data in each of the plurality of groups at the input layer.

33. The system according to claim 27, wherein the plurality of groups applied to the separate processing nodes in the intermediate layer comprise employment stability, residential stability, credit history, debt and income history, and miscellaneous variables.

34. The system according to claim 27, wherein the weight determining means sorts, ranks, and normalizes the weighted connections to values between zero and one.