Parameter-based computer evaluation of user accounts based on user account data stored in one or more databases

• US 10,726,491 B1
• Filed: 05/19/2016
• Issued: 07/28/2020
• Est. Priority Date: 12/28/2015
• Status: Active Grant

• Current Assignee: Plaid Inc.
• Original Assignee: Plaid Inc.
• Inventors: Hockey, William, Rapoport, Sharon
• Primary Examiner(s): Rosen, Elizabeth H
• Assistant Examiner(s): Poe, Kevin T
• Application Number:
  US15/159,714
• Time in Patent Office: 1,531 Days
• Field of Search: 705 35
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06Q 40/02 Banking, e.g. interest calculation or account maintenance credit or loans G06Q40/03
• CPC Class Codes: G06Q 40/08 Insurance
• CPC Class Codes: H04L 63/0272 Virtual private networks
• CPC Class Codes: H04L 63/08 for authentication of entities cryptographic mechanisms or cryptographic arrangements for entity authentication H04L9/32
• CPC Class Codes: H04L 63/1433 Vulnerability analysis

Abstract

A user account evaluation system is disclosed for evaluating risk associated with a user account. The system may obtain user account data associated with many user accounts, select a statistically significant subset of the user accounts, and then process (e.g., to determine types of the user accounts, etc.) and analyze the subset of user accounts to generate a plurality of evaluation models. When a new user account is accessed by the system, user account data may be obtained for the new user account, and the new user account may be evaluated based on the plurality of evaluation models. Accordingly, a plurality of evaluation parameter scores may be generated for the new user account, each of which may indicate an amount of risk associated with the user account. Some embodiments of the present disclosure may include machine learning and/or artificial intelligence methods to improve evaluation of the user accounts.

Claims

1. A computer-implemented method of evaluating a user account based on user account data stored in one or more databases, the computer-implemented method comprising:

by one or more hardware processors executing computer-readable instructions;

storing, in one or more databases, a plurality of user accounts and respective user account data associated with each of the plurality of user accounts, wherein each of the plurality of user accounts is associated with a respective one of a plurality of user account types;

storing, in one or more databases, a plurality of sets of evaluation parameters, wherein each set of evaluation parameters of the plurality of sets of evaluation parameters is associated with a respective one of the plurality of user account types;

selecting a first subset of user accounts of the plurality of user accounts, wherein each user account of the first subset of user accounts is associated with a first user account type;

selecting a first set of evaluation parameters of the plurality of sets of evaluation parameters, wherein the first set of evaluation parameters is associated with the first user account type;

generating a first plurality of normalized evaluation parameter functions by, for each evaluation parameter of the first set of evaluation parameters;

evaluating each user account of the first subset of user accounts by;

accessing, from the one or more databases, user account data associated with the user account; and

analyzing, based on the evaluation parameter, the user account data to determine an evaluation parameter raw value for the user account;

generating a raw distribution based on the evaluation parameter raw values from each of the user accounts of the first subset of user accounts;

generating an evaluation parameter function that estimates a shape of the raw distribution; and

normalizing the evaluation parameter function to generate a normalized evaluation parameter function;

instantiating a simulated instance of a software application, the software application being a first-party application associated with an external user account system, the software application specifically configured to interface via an application programming interface (“

API”

) of the external user account system with computing devices associated with the external user account system, wherein;

the simulated instance of the software application is also configured to interface, via the API of the external user account system, with computing devices associated with the external user account system, and

the simulated instance of the software application is configured to appear to the computing devices associated with the external user account system to be the software application executing on a physical computing device of a customer of the external user account system;

requesting, by the simulated instance of the software application and via the API, new user account data associated with a new user account of a user from at least one of the computing devices associated with the external user account system;

receiving the new user account data from the at least one of the computing devices associated with the external user account system;

determining that the new user account is associated with the first user account type;

generating a plurality of evaluation parameter scores by, for each evaluation parameter of the first set of evaluation parameters;

analyzing, based on the evaluation parameter, the new user account data to determine an evaluation parameter raw value for the new user account; and

determining, based on the normalized evaluation parameter function, of the first plurality of normalized evaluation parameter functions, associated with the evaluation parameter and the evaluation parameter raw value for the new user account, an evaluation parameter score;

receiving, via a standardized API different from the API of the external user account system;

account credentials associated with the new user account; and

a request for the plurality of evaluation parameter scores associated with the new user account; and

providing, via the standardized API different from the API of the external user account system, the plurality of evaluation parameter scores.

2. The computer-implemented method of claim 1 further comprising:

by the one or more hardware processors executing computer-readable instructions;

determining the external user account system associated with the new user account; and

authenticating, using the account credentials, the simulated instance of the software application with the external user account system to establish communication with the external user account system.

3. The computer-implemented method of claim 1, wherein the first set of evaluation parameters includes an overall evaluation parameter, and wherein the computer-implemented method further comprises:

by the one or more hardware processors executing computer-readable instructions;

generating an normalized overall evaluation parameter function by;

for each user account of the first subset of user accounts, determining an overall evaluation parameter raw value by;

selecting, based on the overall evaluation parameter, a subset of the evaluation parameters of the first set of evaluation parameters; and

combining, based on the overall evaluation parameter, evaluation parameter raw values of the user account associated with respective evaluation parameters of the subset of evaluation parameters;

generating an overall raw distribution based on the overall evaluation parameter raw values from each of the user accounts of the first subset of user accounts;

generating an overall evaluation parameter function that estimates a shape of the overall raw distribution; and

normalizing the overall evaluation parameter function to generate a normalized overall evaluation parameter function.

4. The computer-implemented method of claim 3, wherein combining the evaluation parameter raw values comprises at least one of:

summing the evaluation parameter raw values, weighted summing the evaluation parameter raw values, or multiplying the evaluation parameter raw values.

5. The computer-implemented method of claim 4, wherein the overall evaluation parameter is provided by a customer.

6. The computer-implemented method of claim 3 further comprising:

by the one or more hardware processors executing computer-readable instructions;

combining, based on the overall evaluation parameter, evaluation parameter raw values of the new user account associated with respective evaluation parameters of the subset of evaluation parameters to determine an overall evaluation parameter raw value;

determining, based on the normalized overall evaluation parameter function and the overall evaluation parameter raw value for the new user account, an overall evaluation parameter score.

7. The computer-implemented method of claim 1 further comprising:

by the one or more hardware processors executing computer-readable instructions;

obtaining the respective user account data associated with each of the plurality of user accounts from a plurality of external user account systems;

processing user account data items of the respective user account data by;

standardizing a format of the user account data items;

categorizing the user account data items; and

augmenting the user account data items with geolocation information; and

determining, based on the respective user account data, a user account type of the plurality of user account types associated with each of the plurality of user accounts.

8. The computer-implemented method of claim 7, wherein the user account data includes transactions, and wherein the transactions include at least a date and an amount.

9. The computer-implemented method of claim 7, wherein obtaining the respective user account data comprises:

by the one or more hardware processors executing computer-readable instructions;

for each user account of the plurality of user accounts;

receiving account credentials associated with the user account;

determining an external user account system of the plurality of external user account systems associated with the user account;

determining a first-party application configured to interface with the external user account system;

instantiating a simulated instance of the first-party application;

authenticating, using the account credentials, the simulated instance of the first-party application with the external user account system to establish communication with the external user account system; and

accessing, via the virtualized simulated instance of the first-party application, the user account data associated with the user account.

10. The computer-implemented method of claim 1, wherein the first subset of user accounts includes a statistically significant number of user accounts of the plurality of user accounts.

11. The computer-implemented method of claim 10, wherein the statistically significant number comprises at least a number of user accounts, wherein the number comprises at least one of:

100, 1000, 10,000, or 20,000.

12. The computer-implemented method of claim 1, wherein the plurality of user account types include at least one of:

depository, credit, or investment.

13. The computer-implemented method of claim 1, wherein the first set of evaluation parameters includes at least one of:

a number cash advances per month, a number of fee-type transactions per month, a number of late payments per month, a number of foreign fees per month, a number of overdraft transactions per month, a number of manual credit card payments per month, a time between credit card payments, a percent of transactions being in a particular location, a number of transactions starting with a particular digit, a fraction of transactions being foreign fees, or a number of transactions being bank fees.

14. The computer-implemented method of claim 1, wherein generating the raw distribution comprises:

by the one or more hardware processors executing computer-readable instructions;

determining bins into which each of the user accounts of the first subset of user accounts may be placed; and

putting the user accounts of the first subset of user accounts into the determined bins based on the respective evaluation parameter raw values.

15. The computer-implemented method of claim 14, wherein the raw distribution comprises a histogram.

16. The computer-implemented method of claim 1, wherein generating and/or normalizing the evaluation parameter function comprises:

by the one or more hardware processors executing computer-readable instructions;

applying an estimation technique to the raw distribution to estimate a probability density function of the raw distribution.

17. The computer-implemented method of claim 1, wherein the estimation technique comprises kernel density estimation, data clustering, vector quantization, or similar techniques.

18. The computer-implemented method of claim 1, wherein determining the valuation parameter score comprises:

by the one or more hardware processors executing computer-readable instructions;

plugging the evaluation parameter raw value for the new user account into the normalized evaluation parameter function to determine a numerical score between 0 and 1,

wherein a score of or close to 0 indicates a lower risk and a score of or close to 1 indicates a higher risk.

19. The computer-implemented method of claim 1 further comprising:

by the one or more hardware processors executing computer-readable instructions;

selecting a second subset of user accounts of the plurality of user accounts, wherein each user account of the second subset of user accounts is associated with a second user account type;

selecting a second set of evaluation parameters of the plurality of sets of evaluation parameters, wherein the second set of evaluation parameters is associated with the second user account type;

generating a second plurality of normalized evaluation parameter functions by, for each evaluation parameter of the second set of evaluation parameters;

evaluating each user account of the second subset of user accounts by;

accessing, from the one or more databases, user account data associated with the user account; and

analyzing, based on the evaluation parameter, the user account data to determine an evaluation parameter raw value for the user account;

generating a raw distribution based on the evaluation parameter raw values from each of the user accounts of the second subset of user accounts;

generating an evaluation parameter function that estimates a shape of the raw distribution; and

normalizing the evaluation parameter function to generate a normalized evaluation parameter function;

obtaining, from a second external user account system and by instantiating a second simulated instance of a first-party application associated with the second external user account system, second new user account data associated with a second new user account of a user;

determining that the second new user account is associated with the second user account type; and

generating a second plurality of evaluation parameter scores by, for each evaluation parameter of the second set of evaluation parameters;

analyzing, based on the evaluation parameter, the second new user account data to determine an evaluation parameter raw value for the second new user account; and

determining, based on the normalized evaluation parameter function, of the second plurality of normalized evaluation parameter functions, associated with the evaluation parameter and the evaluation parameter raw value for the second new user account, a second evaluation parameter score.