Enhancing Delinquent Debt Collection Using Statistical Models of Debt Historical Information and Account Events

• US 20070156557A1
• Filed: 03/08/2007
• Published: 07/05/2007
• Est. Priority Date: 02/01/2000
• Status: Active Grant

• Current Assignee: Fair Isaac Corporation
• Original Assignee: Fair Isaac Corporation
• Inventors: Drossu, Radu, Martin, Ron, Shao, Min, Cameron, Gordon, Zoldi, Scott, Zhang, Jenny (Guofeng), Shoham, Daniel
• Granted Patent: US 7,536,348 B2
• Time in Patent Office: Days
• US Class Current: 705/35
• CPC Class Codes: G06Q 10/06312 Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 20/102 Bill distribution or payments
• CPC Class Codes: G06Q 40/00 Finance; Insurance; Tax strategies; Processing of corporate or income taxes
• 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 40/08 Insurance

Abstract

A predictive model, for example, a neural network, evaluates individual debt holder accounts and predicts the amount that will be collected on each account based on learned relationships among known variables. The predictive model is generated using historical data of delinquent debt accounts, the collection methods used to collect the debts in the accounts, and the success of the collection methods. In one embodiment, the predictive model is generated using profiles of delinquent debt accounts summarizing patterns of events in the accounts, and the success of the collection effort in each account. In another embodiment, the predictive model includes a mathematical representation of the collector'"'"'s notes created during the collection period for each account.

Claims

1. A computer implemented method of predicting the likelihood of collecting on a delinquent debt on an account, the method comprising the computer implemented steps of:

statistically generating a predictive model of debt collection likelihood using historical data of delinquent debt accounts, specific collection methods used in each debt account, and the success of the collection methods used in each said debt account;

storing a said predictive model;

generating a collector'"'"'s notes model using said historical data of delinquent debt accounts;

storing said statistically generated collectors'"'"' notes model, said statistically generated collectors'"'"' notes model using different types of collectors'"'"' notes'"'"' subject matter determining a collectors'"'"' notes word space;

wherein said different types of collectors'"'"' notes'"'"' subject matter comprises information to determine prior actions taken and circumstances associated with said account;

receiving data of a currently delinquent debt account;

transforming collectors'"'"' notes of said currently delinquent debt account into a mathematical representation thereof and performing any of;

comparing said document mathematical representation against a mathematical representation of said collectors'"'"' notes model, determining a subject matter similarities result based on the comparison, and using said result as input into said predictive model; and

using components of said document mathematical representation as input into said predictive model;

selecting a collection method;

generating a signal indicative of the likelihood of collecting on the currently delinquent debt by applying the data of the currently delinquent debt account and the selected collection method to the predictive model; and

prioritizing collection resource expenditures by using said signal indicative of the likelihood of collecting on the currently delinquent debt for, prioritizing collection resource expenditures.

2. The method of claim 1, wherein the delinquent debt was incurred on any of a credit card, a medical service, on a utility bill, an unpaid check, mail-ordered goods, and an electronic transaction via the Internet, an electronic transaction via the Internet.

3. The method of claim 1, wherein the delinquent debt has been charged-off.

4. The method of claim 1, wherein the collection methods include any of a set of different letters that can be sent to delinquent debtors, different times at which a letter can be sent to delinquent debtors, different phone call approaches, different debt lifecycle stages when phone calls may be made, different debt lifecycle stages at which a letter can be sent to delinquent debtors, different days of the week during which phone calls may be made, different monthly dates during which phone calls may be made, different hours of the day during which phone calls may be made, different collections specialists who may be assigned to work the account, electronic communications that may be made with the debtor, offering debt counseling, debt rescheduling, offering additional credit, changing delinquency financial penalties for an account, offering debt forgiveness, a search for a missing debtor, legal actions, the employment of a collection agency;

and the sale of a debt.

5. The method of claim 1, wherein the historical data include information regarding an account before the account became delinquent, account purchase information, information regarding the Merchant Category Code of purchases on the account, information regarding the amount of account purchases, information regarding account cash transactions, information regarding account payments made, events related to previous collection activities;

and collectors'"'"' notes related to previous collection activities

6. The method of claim 5, wherein the collectors'"'"' notes use and of preformatted codes, and a natural language format.

7. The method of claim 5, further comprising:

transforming the collectors'"'"' notes into a mathematical representation that encodes contextual similarity of terms contained in the collector'"'"'s notes.

8. The method of claim 7, further comprising creating the mathematical representation by any of using a vector model, determining co-occurrence statistics of terms contained in the collector'"'"'s notes, and using context vector methodology.

9. The method of claim 1, further comprising:

segmenting a portfolio of accounts into a plurality of segments; and

providing a different predictive model for each segment.

10. The method of claim 9, wherein each segment is based on any of the time period for which an account has been delinquent, the credit-worthiness of the holder of a delinquent debt account, the type of debt of an account, the history of collection activities for an account, a statistical clustering of accounts having similar characteristics, the amount owed on an account, the collectors'"'"' notes for an account, a determination of whether a debt on an account has been charged-off, and the number of collection agencies that have attempted to collect on the debt account.

11. An apparatus for predicting the likelihood of collecting on a delinquent debt on an account, comprising means for statistically generating a predictive model of debt collection likelihood using historical data of delinquent debt accounts, specific collection methods used in each debt account, and the success of the collection methods used in each said debt account;

a memory for storing a said predictive model;

means for generating a collector'"'"'s notes model using said historical data of delinquent debt accounts;

memory for storing said statistically generated collectors'"'"' notes model, said statistically generated collectors'"'"' notes model using different types of collectors'"'"' notes'"'"' subject matter determining a collectors'"'"' notes word space;

wherein said different types of collectors'"'"' notes'"'"' subject matter comprises information to determine prior actions taken and circumstances associated with said account;

means for receiving data of a currently delinquent debt account;

means for transforming collectors'"'"' notes of said currently delinquent debt account into a mathematical representation thereof and comprising any of;

means for comparing said document mathematical representation against a mathematical representation of said collectors'"'"' notes model, means for determining a subject matter similarities result based on the comparison, and means for using said result as input into said predictive model; and

means for using components of said document mathematical representation as input into said predictive model;

means for selecting a collection method;

means for generating a signal indicative of the likelihood of collecting on the currently delinquent debt by applying the data of the currently delinquent debt account and the selected collection method to the predictive model; and

means for prioritizing collection resource expenditures by using said signal indicative of the likelihood of collecting on the currently delinquent debt for, prioritizing collection resource expenditures.