Method and systems for enhancing modeling for credit risk scores

1. A computer-implemented method for enhancing modeling for credit risk scores, comprising:

identifying and selecting an enhanced performance definition that aligns with a standard performance definition associated with a pool of consumers, such that there is an expanded pool of consumers associated with the enhanced performance definition, comprising:

identifying the standard performance definition, associated with the pool of consumers;

identifying a plurality of potential enhanced performance definitions which provide more information than the standard performance definition;

selecting the enhanced performance definition from the plurality of potential enhanced performance definitions, the enhanced performance definition aligning with the standard performance definition; and

adding a population segment, associated with the enhanced performance definition, to the pool of consumers, thereby creating an expanded pool of consumers;

generating risk nodes based on combinations of ranges of data for selected behavioral dimensions, from the expanded pool of consumers, the act of generating risk nodes utilizing machine learning methods including at least one of bootstrapping and random forest trees to build relationships between data available on the selected behavioral dimensions;

identifying higher value nodes of the generated risk nodes;

converting the identified higher value nodes to attributes; and

generating a scoring model to produce a standard scorecard based on the converted attributes, wherein the standard scorecard generates credit risk scores that are applicable to the expanded pool of consumers; and

utilizing machine learning methods to update the scoring model to improve a predictive performance of the scoring model.

2. The method of claim 1 , wherein the risk nodes generating step performs the steps of:

identifying the ranges of data in the relevant categories from the credit files, for the expanded pool of consumers;

generating risk nodes based on the identified ranges of data;

identifying higher value nodes of the generated risk nodes; and

combining the identified higher value nodes to achieve maximum predictive power.

3. The method of claim 1 , wherein the risk nodes generating step further comprises applying criteria to generate the risk nodes.

4. The method of claim 1 , wherein the risk nodes generating step further includes selecting randomly one relevant category and corresponding identified ranges of data and then generating risk nodes utilizing the at least randomly selected category.

5. The method of claim 1 , wherein the identifying higher value nodes step includes identifying the nodes that give the largest separation of good and bad rates.

6. The method of claim 5 , further comprising combining the identified higher value nodes to achieve maximum predictive power.

7. The method of claim 6 , wherein maximum predictive power is achieved by selecting higher value nodes that include different ranges of data and behavioral dimensions from one another.

8. The method of claim 1 , wherein the converting step comprises combining the identified higher value nodes and key behaviors to generate the attributes.

9. The method of claim 8 , wherein the attributes represent each behavior and each dimension of each behavior.

10. The method of claim 9 , further comprising eliminating attributes that are invalid.

11. A system for enhancing modeling for credit risk scores, comprising: a memory for storing a standard performance definition associated with a pool of consumers; and

a processor for performing the steps of:

identifying and selecting an enhanced performance definition that aligns with the standard performance definition, such that there is an expanded pool of consumers associated with the enhanced performance definition, comprising:

identifying the standard performance definition, associated with the pool of consumers;

identifying a plurality of potential enhanced performance definitions which provide more information than the standard performance definition;

selecting the enhanced performance definition from the plurality of potential enhanced performance definitions, the enhanced performance definition aligning with the standard performance definition; and

adding a population segment, associated with the enhanced performance definition, to the pool of consumers, thereby creating an expanded pool of consumers;

generating risk nodes based on combinations of ranges of data for selected behavioral dimensions, from the expanded pool of consumers, the act of generating risk nodes utilizing machine learning methods including at least one of bootstrapping and random forest trees to build relationships between data available on the selected behavioral dimensions;

identifying higher value nodes of the generated risk nodes;

converting the identified higher value nodes to attributes;

generating a scoring model to produce a standard scorecard based on the converted attributes, wherein the standard scorecard generates credit risk scores that are applicable to the expanded pool of consumers; and

utilizing machine learning methods to update the scoring model to improve a predictive performance of the scoring model.

12. The system of claim 11 , wherein the processor performs the risk nodes generating step by:

identifying the ranges of data in the relevant categories from the credit files, for the expanded pool of consumers;

generating risk nodes based on the identified ranges of data;

identifying higher value nodes of the generated risk nodes; and

combining the identified higher value nodes to achieve maximum predictive power.

13. The system of claim 11 , wherein the risk nodes generating step performed by the processor further comprises applying criteria to generate the risk nodes.

14. The system of claim 11 , wherein the risk nodes generating step performed by the processor further includes selecting randomly one relevant category and corresponding identified ranges of data and then generating risk nodes utilizing the at least randomly selected category.

15. The system of claim 11 , wherein the identifying higher value nodes step performed by the processor includes identifying the nodes that give the largest separation of good and bad rates.

16. The system of claim 15 , wherein the processor further comprising performing the step of combining the identified higher value nodes to achieve maximum predictive power.

17. The system of claim 16 , wherein maximum predictive power is achieved by selecting higher value nodes that include different ranges of data and behavioral dimensions from one another.

18. The system of claim 11 , wherein the converting step performed by the processor comprises combining the identified higher value nodes and key behaviors to generate the attributes.

19. The system of claim 18 , wherein the attributes represent each behavior and each dimension of each behavior.

20. The system of claim 19 , wherein the processor further comprising the step of eliminating attributes that are invalid.

21. A non-transitory computer readable medium with computer executable instructions embodied thereon for enhancing modeling for credit risk score, the computer executable instructions causing a computer to perform the process of:

identifying and selecting an enhanced performance definition that aligns with a standard performance definition associated with a pool of consumers, such that there is an expanded pool of consumers associated with the enhanced performance definition, comprising:

identifying the standard performance definition, associated with the pool of consumers;

identifying a plurality of potential enhanced performance definitions which provide more information than the standard performance definition;

selecting the enhanced performance definition from the plurality of potential enhanced performance definitions, the enhanced performance definition aligning with the standard performance definition; and

generating risk nodes based on combinations of ranges of data for selected behavioral dimensions, from the expanded pool of consumers, the act of generating risk nodes utilizing machine learning methods including at least one of bootstrapping and random forest trees to build relationships between data available on the selected behavioral dimensions;

generating risk nodes based on combinations of ranges of data, from the expanded pool of consumers;

identifying higher value nodes of the generated risk nodes;

converting the identified higher value nodes to attributes; and

generating a scoring model to produce a standard scorecard based on the converted attributes, wherein the standard scorecard generates credit risk scores that are applicable to the expanded pool of consumers; and

utilizing machine learning methods to update the scoring model to improve a predictive performance of the scoring model.