Dynamically retraining a prediction model based on real time transaction data

Patent metadata

Field	Value
Publication number	US-9449344-B2
Application number	US-201314138135-A
Country	US
Kind code	B2
Filing date	Dec 23, 2013
Priority date	Dec 23, 2013
Publication date	Sep 20, 2016
Grant date	Sep 20, 2016

Abstract

Various embodiments of systems and methods to dynamically retrain prediction models based on real time transaction data are described herein. In one aspect, real time application data and status data associated with an entity are obtained. The obtained application data is inputted to a prediction model to produce an assessment of a risk. The obtained status data with the assessed risk are compared. When the obtained payment status data does not match the determined risk, the prediction model is retrained.

First claim

What is claimed is:

1. A non-transitory computer-readable medium storing instructions, which when executed cause a computer system to perform operations comprising:
obtain, in real time, application data and status data associated with an entity;
input the obtained real time application data to a prediction model to produce an assessment of a risk, wherein the prediction model is initially trained based on historical application data and corresponding status data;
compare the obtained real time status data with the assessed risk; and
retrain the prediction model based on the obtained real time application data and the status data upon determining that the obtained real time status data does not match the assessed risk, wherein retraining the prediction model comprises adjusting hidden node weights and output node weights of the prediction model based on the obtained real time application data.

2. The non-transitory computer-readable medium of claim 1 , wherein the real time application data comprises a plurality of parameters and corresponding values affecting the risk.

3. The non-transitory computer-readable medium of claim 2 , wherein the plurality of parameters are interdependent.

4. The non-transitory computer-readable medium of claim 1 , wherein the real time status data comprises status of deferral of payment associated with the real time application data.

5. The non-transitory computer-readable medium of claim 1 , wherein the prediction model comprises an artificial neural network.

6. The non-transitory computer-readable medium of claim 1 , wherein the historical application data and the corresponding status data is associated with a plurality of entities.

7. The non-transitory computer-readable medium of claim 1 , wherein retraining the prediction model comprises:
receiving the obtained real time application data;
adjusting the hidden node weights and the output node weights of the prediction model based on the obtained real time application data;
computing output nodes of the prediction model based on the initialized hidden node weights and the output node weights;
comparing whether the output nodes match with the obtained real time status data; and
retraining the prediction model by reinitializing the hidden node weights and the output node weights, and computing the output nodes until the output nodes match with the obtained real time status data.

8. A computer implemented method to dynamically retrain a prediction model based on real time transaction data using a computer, the method comprising:
obtaining, in real time, application data and status data associated with an entity;
inputting the obtained real time application data to the prediction model to produce an assessment of a risk, wherein the prediction model is initially trained based on historical application data and corresponding status data;
the computer, comparing the obtained real time status data with the assessed risk; and
the computer, retraining the prediction model based on the obtained real time application data and the status data upon determining that the obtained real time status data does not match the assessed risk, wherein retraining the prediction model comprises adjusting hidden node weights and output node weights of the prediction model based on the obtained real time application data.

9. The computer implemented method of claim 8 , wherein the real time application data comprises a plurality of parameters and corresponding values affecting the risk.

10. The computer implemented method of claim 9 , wherein the plurality of parameters are interdependent.

11. The computer implemented method of claim 8 , wherein the real time status data comprises status of deferral of payment associated with the real time application data.

12. The computer implemented method of claim 8 , wherein the prediction model comprises an artificial neural network.

13. The computer implemented method of claim 8 , wherein the historical application data and the corresponding status data is associated with a plurality of entities.

14. The computer implemented method of claim 8 , wherein retraining the prediction model comprises:
receiving the obtained real time application data;
adjusting the hidden node weights and the output node weights of the prediction model based on the obtained real time application data;
computing output nodes of the prediction model based on the initialized hidden node weights and the output node weights;
comparing whether the output nodes match with the obtained real time status data; and
retraining the prediction model by reinitializing the hidden node weights and the output node weights, and computing the output nodes until the output nodes match with the obtained real time status data.

15. A computer system to dynamically retrain a prediction model based on real time transaction data, the computer system comprising:
at least one processor; and
one or more memory devices communicative with the at least one processor, wherein the one or more memory devices store instructions to:
obtain, in real time, application data and status data associated with an entity;
input the obtained real time application data to the prediction model to produce an assessment of a risk, wherein the prediction model is initially trained based on historical application data and corresponding status data;
compare the obtained real time status data with the assessed risk; and
retrain the prediction model using the obtained real time application data and the status data upon determining that the obtained real time status data does not match the assessed risk, wherein retraining the prediction model comprises adjusting hidden node weights and output node weights of the prediction model based on the obtained real time application data.

16. The computer system of claim 15 , wherein the real time application data comprises a plurality of parameters and corresponding values affecting the risk.

17. The computer system of claim 16 , wherein the plurality of parameters are interdependent.

18. The computer system of claim 15 , wherein the real time status data comprises status of deferral of payment.

19. The computer system of claim 15 , wherein the prediction model comprises an artificial neural network.

20. The computer system of claim 15 , wherein retraining the prediction model comprises:
receiving the obtained real time application data;
adjusting the hidden node weights and the output node weights of the prediction model based on the obtained application data;
computing output nodes of the prediction model based on the initialized hidden node weights and the output node weights;
comparing whether the output nodes match with the obtained real time status data; and
retraining the prediction model by reinitializing the hidden node weights and the output node weights, and computing the output nodes until the output nodes match with the obtained real time status data.

Assignees

• Deshpande Raghavendra
• Monteiro Vivek John
• A R Shruthi
• Sap Se

Inventors

• Deshpande Raghavendra
• Monteiro Vivek John
• A R Shruthi

Classifications

• G06Q40/03

  Credit; Loans; Processing thereof · CPC title

• G06N3/084 Primary

  Backpropagation, e.g. using gradient descent · CPC title

• G06Q10/0635

  Risk analysis of enterprise or organisation activities · CPC title

• G06Q40/00 Primary

  Finance; Insurance; Tax strategies; Processing of corporate or income taxes · CPC title

• G06Q20/4016

  involving fraud or risk level assessment in transaction processing · CPC title