CN-122029553-A - Modifying a predictive model based on qualitative information

Abstract

Techniques are provided for modifying a predictive model. For example, one or more embodiments described herein may include a modeling system that may include a memory that may store computer-executable components. The modeling system may also include a processor operatively coupled to the memory and may execute the computer-executable components stored in the memory. The computer-executable components may include an identification component that can identify qualitative information related to a predictive model of operation of the system, wherein the predictive model is generated from data obtained from operation of the system. The computer-executable components may include a modification component that may modify the predictive model based on intervention information that is generated based on qualitative information, wherein the intervention information is associated with a prediction of the predictive model.

Inventors

• S. MUKHERJEE
• C. Kamanchi
• P. Damama
• GIATTI ANNA
• K. Sampat

Assignees

• 国际商业机器公司

Dates

Publication Date

20260512

Application Date

20241014

Priority Date

20231020

Claims (20)

1. 1. A modeling system, comprising: a memory storing computer-executable components, and A processor operatively coupled to the memory, the processor executing the computer-executable components stored in the memory, wherein the computer-executable components comprise: An identification component that identifies qualitative information related to a predictive model of operation of a system, wherein the predictive model is generated from a dataset obtained from the operation of the system, and A modification component that modifies the predictive model based on intervention information generated based on the qualitative information, wherein the intervention information is associated with a prediction of the predictive model.
2. 2. The modeling system of claim 1, wherein the qualitative information includes an indication that the system is operating differently than has been modeled by the predictive model.
3. 3. The modeling system of claim 2, wherein the indication includes qualitative monotonic information regarding the prediction and the operation of the system.
4. 4. The modeling system of any of the preceding claims, wherein the computer-executable component further comprises a range component that identifies a range of input values that are related to the qualitative information, wherein the intervention information is based on the range.
5. 5. The modeling system of claim 4, wherein the range is identified based on the predictive model that includes a lesser amount of predictive data for the range than other input values not included in the range.
6. 6. The modeling system of claim 5, wherein the predictive model includes the lesser amount of predictive data for the range based on a determination that testing the system based on the range is predicted to be destructive to the system.
7. 7. A modeling system according to any preceding claim wherein the predictive model comprises a data driven predictive model of the system.
8. 8. The modeling system of any of the preceding claims, wherein the qualitative information converted into the intervention information includes the qualitative information converted into a declarative language.
9. 9. A computer-implemented method, comprising: Identifying, by a device operatively coupled to a processor, qualitative information related to a predictive model of operation of a system, wherein the predictive model is generated from a dataset obtained from the operation of the system, and Modifying, by the device, the predictive model based on intervention information generated based on the qualitative information, wherein the intervention information is associated with a prediction of the predictive model.
10. 10. The computer-implemented method of claim 9, wherein the qualitative information includes an indication that the system has operated differently than modeled by the predictive model.
11. 11. The computer-implemented method of claim 10, wherein the indication comprises qualitative monotonic information regarding the prediction and the operation of the system.
12. 12. The computer-implemented method of any of the preceding claims 9 to 11, further comprising identifying a range of input values related to the qualitative information, wherein the intervention information is based on the range.
13. 13. The computer-implemented method of claim 12, wherein the range is identified based on the predictive model that includes a lesser amount of predicted data for the range than other input values not included in the range.
14. 14. The computer-implemented method of claim 13, wherein the predictive model includes the lesser amount of predictive data for the range based on a determination that testing the system based on the range is predicted to be destructive to the system.
15. 15. The computer-implemented method of any of the preceding claims 9 to 14, wherein the predictive model comprises a data-driven predictive model of the system.
16. 16. The computer-implemented method of any of the preceding claims 9 to 15, wherein the qualitative information converted into the intervention information comprises qualitative information converted into a declarative language.
17. 17. A computer program product for modifying a predictive model of operation of a system, the computer program product comprising a computer-readable storage medium having program instructions embodied therewith, the program instructions being executable by a processor to cause the processor to: Identifying qualitative information related to the predictive model, wherein the predictive model is generated from data obtained from the operation of the system, and The predictive model is modified based on intervention information generated based on the qualitative information, wherein the intervention information is associated with a prediction of the predictive model.
18. 18. The computer program product of claim 17, wherein the qualitative information includes an indication that the system is operating differently than has been modeled by the predictive model.
19. 19. The computer program product of claim 18, wherein the indication comprises qualitative monotonic information regarding the prediction and the operation of the system.
20. 20. The computer program product of any of the preceding claims 17 to 19, further comprising identifying a range of input values related to the qualitative information, wherein the intervention information is based on the range.

Description

Modifying a predictive model based on qualitative information Background One or more embodiments relate to predictive models, and more particularly to modifying predictive models. Disclosure of Invention The following presents a simplified summary in order to provide a basic understanding of one or more embodiments of the disclosure. This summary is not intended to identify key or critical elements or to delineate any scope of the particular embodiments or any scope of the claims. Its sole purpose is to present the concepts in a simplified form as a prelude to the more detailed description that is presented later. In one or more embodiments described herein are systems, computer-implemented methods, apparatuses, and/or computer program products that may utilize intermediate data sets to generate a predictive model. According to some embodiments described herein, a modeling system is provided. The modeling system may include a memory that stores computer-executable components. The system may also include a processor operatively coupled to the memory, the processor may execute the computer-executable components stored in the memory. The computer-executable components may include an identification component that may identify qualitative information related to a predictive model of operation of the system, wherein the predictive model is generated from a dataset obtained from operation of the system. The computer-executable components may also include a modification component that can modify the predictive model based on intervention information that is generated based on the qualitative information, wherein the intervention information is associated with a prediction of the predictive model. In additional or alternative embodiments, the qualitative information may include an indication that the system is operating differently than the system that has been modeled by the predictive model. In additional or alternative embodiments, the indication may include qualitative monotonic information regarding the operation of the system. In additional or alternative embodiments, the computer-executable components may further include a range component that identifies a range of input values associated with the qualitative information, wherein the intervention information is based on the range. In additional or alternative embodiments, a range may be identified based on a predictive model that includes a smaller amount of predictive data for the range than other input values not included in the range, e.g., also referred to as "blind spots" discussed below. In additional or alternative embodiments, the predictive model may include less predictive data for a range based on a determination that the range test system is predicted to be destructive to the system. In additional or alternative embodiments, the qualitative information converted into intervention information may include qualitative information converted into a declarative language. According to one or more example embodiments, a computer-implemented method is provided. The computer-implemented method may identify qualitative information related to a predictive model of operation of the system, wherein the predictive model is generated from a dataset obtained from operation of the system, and may further include modifying the predictive model based on intervention information, wherein the intervention information is generated based on the qualitative information, wherein the intervention information is associated with a prediction of the predictive model. In additional or alternative embodiments, the qualitative information identified by the computer-implemented method may include an indication that the system is operating differently than the system that has been modeled by the predictive model. In additional or alternative embodiments, the indications utilized by the computer-implemented method may include qualitative monotonic information regarding predictions and operations of the system. In additional or alternative embodiments, the method may further include identifying a range of input values associated with the qualitative information, wherein the intervention information is based on the range. In additional or alternative embodiments, a range may be identified based on a predictive model that includes a smaller amount of predictive data for the range than other input values not included in the range. In additional or alternative embodiments, the predictive model may include a smaller amount of predictive data for the range based on a determination that the range test system is predicted to be destructive to the system. In additional or alternative embodiments, the predictive model may include a data-driven predictive model of the system. In additional or alternative embodiments, the qualitative information converted to intervention information by the computer-implemented method may include qualitative information converted to a declarative language. Accord