Building resilient models to address dynamic customer data use rights

1. A method of building a decision or prediction model used for analyzing and scoring behavioral transactions, wherein a customer dataset in a model development store used to build an original model is subject to a data right usage withdrawal, the original model having coverage over the customer dataset, the method comprising:

extracting, by one or more processors using data sampling, a portion of the customer dataset to generate a model surrogate dataset that has a distribution of values with a first degree of similarity to the distribution of values of the customer dataset;

discretizing, by the one or more processors, vectors present in both the model surrogate dataset and the customer dataset;

receiving, by the one or more processors, data representing the data right usage withdrawal from the customer dataset;

determining, by the one or more processors, a depletion of the model surrogate dataset according to the data right usage withdrawal;

computing, by the one or more processors, an estimated mean time to coverage failure of the original model based on the depletion of the model surrogate dataset according to the data right usage withdrawal; and

tracking, by one or more processors, a consent validity of a customer associated with the customer dataset, the consent validity representing the customer's continued consent validity for use in the original model.

2. The method in accordance with claim 1 , wherein computing an estimated mean time to coverage failure further includes measuring, by the one or more processors, a mean and an expected model validity failure time based on the depletion of the model surrogate data according to the data right usage withdrawal.

3. The method in accordance with claim 1 , further comprising storing, by the one or more processors, the customer dataset as a set of key values in an in-memory database, at least one key value having a primary key to uniquely identify a customer of the customer dataset.

4. The method in accordance with claim 3 , further comprising generating, by the one or more processors, a secondary key associated with at least one primary key, the secondary key corresponding to a transaction data point for the customer of the customer dataset.

5. The method in accordance with claim 3 , further comprising applying, by the one or more processors, a one-way hash function on the primary key to generate a hashed value to identify the customer of the customer dataset.

6. The method in accordance with claim 1 , further comprising:

replacing, by one or more processors, customer data points removed due to removal of consent for use in the original model and replacing with similar surrogate data points from the model surrogate dataset.

7. The method in accordance with claim 2 , further comprising comparing, by the one or more processors, at least one customer subject to the data right usage withdrawal with the model surrogate dataset to determine the estimated mean time to coverage failure of the original model.

8. A system for building a decision or prediction model used for analyzing and scoring behavioral transactions, wherein a customer dataset in a model development store used to build an original model is subject to a data right usage withdrawal, the original model having coverage over the customer dataset, the system comprising computer hardware configured to perform operations comprising:

extracting, using data sampling, a portion of the customer dataset to generate a model surrogate dataset that has a distribution of values with a first degree of similarity to the distribution of values of the customer dataset;

discretizing vectors present in both the model surrogate dataset and the customer dataset;

receiving data representing the data right usage withdrawal from the customer dataset;

determining a depletion of the model surrogate dataset according to the data right usage withdrawal; and

computing an estimated mean time to coverage failure of the original model based on the depletion of the model surrogate dataset according to the data right usage withdrawal.

9. The system in accordance with claim 8 , wherein the operations further comprise computing an estimated mean time to coverage failure further includes measuring a mean and an expected model validity failure time based on the depletion of the model surrogate data according to the data right usage withdrawal.

10. The system in accordance with claim 8 , wherein the operations further comprise storing the customer dataset as a set of key values in an in-memory database, at least one key value having a primary key to uniquely identify a customer of the customer dataset.

11. The system in accordance with claim 10 , wherein the operations further comprise generating a secondary key associated with at least one primary key, the secondary key corresponding to a transaction data point for the customer of the customer dataset.

12. The system in accordance with claim 10 , further comprising applying, by the one or more processors, a one-way hash function on the primary key to generate a hashed value to identify the customer of the customer dataset.

13. The system in accordance with claim 8 , wherein the operations further comprise:

tracking a consent validity of a customer associated with the customer dataset, the consent validity representing the customer's continued consent validity for use in the original model; and

replacing customer data points removed due to removal of consent for use in the original model and replacing with similar surrogate data points from the model surrogate dataset.

14. The system in accordance with claim 9 , wherein the operations further comprise comparing at least one customer subject to the data right usage withdrawal with the model surrogate dataset to determine the estimated mean time to coverage failure of the original model.

15. A system comprising:

a programmable processor; and

a machine-readable medium storing instructions that, when executed by the processor, cause the at least one programmable processor to perform operations comprising:

extract, using data sampling, a portion of the customer dataset to generate a model surrogate dataset;

discretize vectors present in both the model surrogate dataset and the customer dataset;

receive data representing the data right usage withdrawal from the customer dataset;

determine a depletion of the model surrogate dataset according to the data right usage withdrawal;

compute an estimated mean time to coverage failure of the original model based on the depletion of the model surrogate dataset according to the data right usage withdrawal; and

track, by one or more processors, a consent validity of a customer associated with the customer dataset, the consent validity representing the customer's continued consent validity for use in the original model.

16. The system in accordance with claim 15 , wherein the operations further comprise compute an estimated mean time to coverage failure further includes measuring a mean and an expected model validity failure time based on the depletion of the model surrogate data according to the data right usage withdrawal.

17. The system in accordance with claim 15 , wherein the operations further comprise store the customer dataset as a set of key values in an in-memory database, at least one key value having a primary key to uniquely identify a customer of the customer dataset.

18. The system in accordance with claim 17 , wherein the operations further comprise generate a secondary key associated with at least one primary key, the secondary key corresponding to a transaction data point for the customer of the customer dataset.

19. The system in accordance with claim 17 , further comprising applying, by the one or more processors, a one-way hash function on the primary key to generate a hashed value to identify the customer of the customer dataset.

20. The system in accordance with claim 15 , wherein the operations further comprise:

replace customer data points removed due to removal of consent for use in the original model and replacing with similar surrogate data points from the model surrogate dataset.

21. The system in accordance with claim 16 , wherein the operations further comprise compare at least one customer subject to the data right usage withdrawal with the model surrogate dataset to determine the estimated mean time to coverage failure of the original model.