Technology for building and managing data models

Patent metadata

Field	Value
Publication number	US-12243106-B2
Application number	US-202318221023-A
Country	US
Kind code	B2
Filing date	Jul 12, 2023
Priority date	Nov 21, 2017
Publication date	Mar 4, 2025
Grant date	Mar 4, 2025

Abstract

Techniques for building and managing data models are provided. According to certain aspects, systems and methods may enable a user to input parameters associated with building one or more data models, including parameters associated with sampling, binning, and other factors. The systems and methods may automatically generate program code that corresponds to the inputted parameters and display the program code for review by the user. The systems and methods may build the data models and generate charts and plots depicting aspects of the data models. Additionally, the systems and methods may combine data models and select champion data models.

First claim

What is claimed:

1. A computer-implemented method in a computing device of enabling the management of data models, the method comprising:
generating, by a computer processor, a model build partition, including enabling a user to input, via a user interface:
a storage location where a modeling output is to be stored,
a set of variables to be binned,
a set of identifications for (i) at least one of a training dataset and a validation dataset, and (ii) modeling data, and
a set of selections associated with (i) whether to generate the modeling output using the training dataset and the validation dataset, or using the training dataset, (ii) a model iteration identification, and (iii) a set of model effects;

generating, by the processor, the modeling output according to the model build partition; and
displaying, in the user interface, a set of results associated with generating the modeling output, the set of results including: (a) a set of model level results, and (b) a set of variable level results.

2. The computer-implemented method of claim 1 , wherein enabling the user to input the set of variables to be binned comprises:
enabling the user to input, for each of the set of variables, (i) a binning technique, (ii) a number of bins, and/or (iii) a binned value.

3. The computer-implemented method of claim 1 , wherein enabling the user to input the set of identifications for the modeling data comprises:
enabling the user to input (i) a model type, (ii) a distribution, (iii) a link function, and/or (iv) a unique identifier.

4. The computer-implemented method of claim 1 , wherein enabling the user to input the set of identifications for the modeling data comprises:
enabling the user to input a claim count variable and/or an exposure variable.

5. The computer-implemented method of claim 1 , wherein displaying the set of model level results comprises:
displaying, in the user interface, a set of prediction statistics.

6. The computer-implemented method of claim 1 , wherein displaying the set of variable level results comprises:
displaying, in the user interface, a set of main effects or a set of interaction relativity plots.

7. The computer-implemented method of claim 1 , wherein the modeling output comprises a first model output and a second model output, and wherein the method further comprises:
combining the first model output and the second model output using either an additive technique or a multiplicative technique.

8. The computer-implemented method of claim 7 , further comprising, after combining the first model output and the second model output:
selecting a champion model.

9. A system for enabling the management of data models, comprising:
a user interface;
a memory storing a set of computer-executable instructions; and
a processor interfaced with the user interface and the memory, and configured to execute the computer-executable instructions to cause the processor to:
generate a model build partition, including enabling a user to input, via the user interface:
a storage location where a modeling output is to be stored,
a set of identifications for (i) at least one of a training dataset and a validation dataset, and (ii) modeling data, and
a set of selections associated with (i) whether to generate the modeling output using the training dataset and the validation dataset, or using the training dataset, (ii) a model iteration identification, and (iii) a set of model effects, generate the modeling output according to the model build partition, and

cause the user interface to display a set of results associated with generating the modeling output, the set of results including: (a) a set model level results, and (b) a set of variable level results.

10. The system of claim 9 , wherein to enable the user to input the set of variables to be binned, the processor is configured to:
enable the user to input, for each of the set of variables, (i) a binning technique, (ii) a number of bins, and/or (iii) a binned value.

11. The system of claim 9 , wherein to enable the user to input the set of identifications for the modeling data, the processor is configured to:
enable the user to input (i) a model type, (ii) a distribution, (iii) a link function, and/or (iv) a unique identifier.

12. The system of claim 9 , wherein to enable the user to input the set of identifications for the modeling data, the processor is configured to:
enable the user to input a claim count variable and/or an exposure variable.

13. The system of claim 9 , wherein to cause the user interface to display the set of model level results, the processor is configured to:
cause the user interface to display a set of prediction statistics.

14. The system of claim 9 , wherein to cause the user interface to display the set of variable level results, the processor is configured to:
cause the user interface to display a set of main effects or a set of interaction relativity plots.

15. The system of claim 9 , wherein the modeling output comprises a first model output and a second model output, and wherein the processor is further configured to:
combine the first model output and the second model output using either an additive technique or a multiplicative technique.

16. The system of claim 15 , wherein the processor is further configured to, after combining the first model output and the second model output:
select a champion model.

17. A computer program product comprising a non-transitory computer readable medium storing a set of computer instructions executable by at least one processor to:
generate a model build partition, including enabling a user to input, via a user interface:
a storage location where a modeling output is to be stored,
a set of variables to be binned,
a set of identifications for (i) at least one of a training dataset and a validation dataset, and (ii) modeling data, and
a set of selections associated with (i) whether to generate the modeling output using the training dataset and the validation dataset, or using the training dataset, (ii) a model iteration identification, and (iii) a set of model effects;

generate the modeling output according to the model build partition; and
display, in the user interface, a set of results associated with generating the modeling output, the set of results including: (a) a set of model level results, and (b) a set of variable level results.

18. The computer program product of claim 17 , wherein to enable the user to input the set of variables to be binned, the at least one processor executes the set of computer instructions to:
enable the user to input, for each of the set of variables, (i) a binning technique, (ii) a number of bins, and/or (iii) a binned value.

19. The computer program product of claim 17 , wherein to enable the user to input the set of identifications for the modeling data, the at least one processor executes the set of computer instructions to:
enable the user to input (i) a model type, (ii) a distribution, (iii) a link function, and/or (iv) a unique identifier.

20. The computer program product of claim 17 , wherein to enable the user to input the set of identifications for the modeling data, the at least one processor executes the set of computer instructions to:
enable the user to input a claim count variable and/or an exposure variable.

Assignees

• State Farm Mutual Automobile Insurance Co

Inventors

• Wu Weixin
• Moon Philip Sangpil
• Farris Scott

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06F18/2193

  based on specific statistical tests · CPC title

• G06F18/2163

  Partitioning the feature space · CPC title

• G06F18/40

  Software arrangements specially adapted for pattern recognition, e.g. user interfaces or toolboxes therefor · CPC title

• G06N20/00

  Machine learning · CPC title