System and method for probate prediction

• US 8,170,895 B1
• Filed: 01/16/2009
• Issued: 05/01/2012
• Est. Priority Date: 01/16/2009
• Status: Active Grant

• Current Assignee: Forte LLC
• Original Assignee: Forte LLC
• Inventors: Larsen, Andrew M.
• Primary Examiner(s): Gart, Matthew S
• Assistant Examiner(s): IWARERE, OLUSEYE
• Application Number:
  US12/355,449
• Time in Patent Office: 1,201 Days
• Field of Search: 705/1.1, 705/7.11
• US Class Current: 705/7.11
• CPC Class Codes: G06Q 10/063 Operations research, analysis or management
• CPC Class Codes: G06Q 50/186 Estate planning

Abstract

A computer assisted method includes selecting a sample of accounts from a historical database of accounts corresponding to deceased debtors, The sample indicates whether or not an estate was found for each deceased debtor. A comprehensive set of credit related variables corresponding to the accounts is obtained from a first source. A comprehensive set of demographic related variables corresponding to the accounts is obtained from a second source. The variables are mapped to the accounts and a computer executable model is created by identifying a subset of variables from the sets of variables and using the subset of variables and sample of accounts such that the model provides a prediction of whether or not an estate exists for a given deceased debtor.

Claims

1. A computer assisted method comprising:

selecting a sample of accounts from a historical database of accounts corresponding to deceased debtors, the sample indicating whether or not an estate was found for each deceased debtor;

obtaining a comprehensive set of credit related variables corresponding to the accounts from a first source;

obtaining a comprehensive set of demographic related variables corresponding to the accounts from a second source;

obtaining individual characteristic variables for each deceased debtor;

mapping the credit related variables, demographic related variables, and the individual characteristic variable variables to the accounts via the computer; and

creating a computer executable model via the computer by identifying a subset of variables from the sets of variables and using the subset of variables and sample of accounts such that the model provides a prediction of whether or not an estate exists for a given deceased debtor.

2. The method of claim 1 wherein selecting a sample of accounts further comprises keeping a sample slice out of the sample for use in testing the model.

3. The method of claim 2 wherein the sample of accounts is selected from a group of accounts processed in accordance with a constant method of identifying whether or not the deceased debtors had estates.

4. The method of claim 1 wherein the first source comprises at least one credit profile source.

5. The method of claim 1 wherein the second source comprises at least one census database.

6. The method of claim 5 wherein the demographic data comprises at least a zipcode and income information.

7. The method of claim 1 wherein the historical database of accounts comprises further variables used to create the model.

8. The method of claim 1 wherein the model comprises a probit model.

9. The method of claim 8 wherein the probit model is created using a computer assisted manual intuitive model.

10. The method of claim 8 wherein the probit model is created using a stepwise logistical process.

11. The method of claim 10 wherein the stepwise logistical process comprises;

running the model from the sets of variables;

removing variables that appear to be duplicative;

running the model without the duplicative variables; and

repeating the removing duplicative variables and running of the model in an empirical process to identify a preliminary subset of variables.

12. The method of claim 11 and further comprising:

forcing the subset of variables to be in the model and iteratively;

selecting multiple variables not in the subset to run the model with the subset of variables such that the model identifies further significant variables;

adding the further significant variables to the subset of variables; and

removing selected variables that appear to be duplicative from the subset of variables.

13. The method of claim 1 wherein the model produces a z score for a selected deceased debtor.

14. The method of claim 13 wherein the z score is representative of the likelihood that an account will have an estate, and a correspondingly increased chance of collecting a debt associated with the deceased debtor.

15. The method of claim 14 wherein the z score is transformed using the standard normal distribution and a scaling equation to produce a probate prediction score.

16. A computer executable model stored on a computer readable storage device, the model comprising:

a set of variables selected from credit and demographic data sources, the variables selected from a method comprising;

selecting a sample of accounts from a historical database of accounts corresponding to deceased debtors, the sample indicating whether or not an estate was found for each deceased debtor;

obtaining a comprehensive set of credit related variables corresponding to the accounts from a first source;

obtaining a comprehensive set of demographic related variables corresponding to the accounts from a second source;

mapping the sets of variables to the accounts; and

creating a computer executable model by identifying a subset of variables from the sets of variables and using the subset of variables and sample of accounts such that the model provides a prediction of whether or not an estate exists for a given deceased debtor.

17. A computer executable model of claim 16 wherein the code comprises a probit model created using a computer assisted manual intuitive model.

18. A computer implemented method comprising:

associating credit and demographic data corresponding to a deceased debtor to a deceased debtor account stored on a computer readable storage device;

running on a programmed computer, a statistical model trained on historical deceased debtor data against the deceased debtor account; and

identifying via the programmed computer, a likelihood that there is an estate corresponding to the deceased debtor account.

19. The method of claim 18 and further comprising performing the method on multiple debtor accounts and identify priority debtor accounts that are more likely to have an estate and thus an increased chance of collection to obtain higher collection yields.

20. The method of claim 18 wherein a low probate prediction score identifies an account as a candidate for reduced collection efforts.