Cohort half life forecasting combination from a confident jury

• US 9,342,790 B1
• Filed: 01/07/2015
• Issued: 05/17/2016
• Est. Priority Date: 01/07/2015
• Status: Expired due to Fees

• Current Assignee: Kyndryl Incorporated (Kyndryl Holdings, Inc.)
• Original Assignee: International Business Machines Corporation
• Inventors: Baughman, Aaron K., Eggenberger, Christian, Malkin, Peter K., Marzorati, Mauro
• Primary Examiner(s): Holmes, Michael B
• Application Number:
  US14/591,403
• Time in Patent Office: 496 Days
• Field of Search: 706/52
• US Class Current: 1/1
• CPC Class Codes: G06N 5/043 Distributed expert systems; Blackboards
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A forecasting cohort includes a first set of forecasting algorithms and a second set of forecasting algorithms. An initial confidence level and a half-life of each of the first set of forecasting algorithms and the second set of forecasting algorithms are determined. A half-life weight for each of the first set of forecasting algorithms and the second set of forecasting algorithms at a subsequent time are determined, such that the half-life weights decrease an effect of a forecasting algorithm as time elapses. A combined confidence level of the forecasting cohort at the subsequent time is determined and used to adjust resource usage.

Claims

1. A method of implementing a forecasting cohort, the method comprising:

generating, by one or more processors, a forecasting cohort of forecasting algorithms, wherein the forecasting cohort comprises a first set of forecasting algorithms of a first type of forecasting algorithm, and wherein the forecasting cohort further comprises a second set of forecasting algorithms of a second type of forecasting algorithm;

determining, by one or more processors, an initial confidence level and a half-life of each of the first set of forecasting algorithms and the second set of forecasting algorithms, wherein the initial confidence level describes an accuracy level of each of the forecasting algorithms at an initial time in predicting a condition, and wherein the half-life describes a subsequent time at which a forecasting algorithm has reached half of its forecast horizon;

determining, by one or more processors, a half-life weight for each of the first set of forecasting algorithms and the second set of forecasting algorithms at a subsequent time that is subsequent to the initial time, wherein half-life weights decrease an effect of a forecasting algorithm as time elapses;

determining, by one or more processors, a combined confidence level of the forecasting cohort at the subsequent time, wherein the combined confidence level is based on the initial confidence level and the half-life weight of each of the first set of forecasting algorithms and the second set of forecasting algorithms; and

utilizing, by one or more processors, the combined confidence level of the forecasting cohort at the subsequent time to adjust resource usage.

2. The method of claim 1, further comprising:

receiving, by one or more processors, sensor readings from a device, wherein the sensor readings describe current conditions of the device, and wherein the device includes resources;

predicting, by one or more processors, future usage of the resources in the device based on the combined confidence level of the forecasting cohort; and

reallocating, by one or more processors, the resources used by the device based on the predicted future usage.

3. The method of claim 2, wherein the resources used by the device are software resources.

4. The method of claim 2, wherein the resources used by the device are hardware resources.

5. The method of claim 2, wherein reallocated resources are in the device.

6. The method of claim 2, wherein the sensor readings are received at the initial time.

7. The method of claim 2, wherein the sensor readings are received at the subsequent time.

8. The method of claim 1, wherein the half-life weight of a forecasting algorithm is:

9. The method of claim 1, wherein the resource usage is predicted at the initial time.

10. The method of claim 1, wherein the resource usage is predicted at the subsequent time.

11. A computer program product for implementing a forecasting cohort, the computer program product comprising a computer readable storage medium having program code embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, and wherein the program code is readable and executable by a processor to perform a method comprising:

generating a forecasting cohort of forecasting algorithms, wherein the forecasting cohort comprises a first set of forecasting algorithms of a first type of forecasting algorithm, and wherein the forecasting cohort further comprises a second set of forecasting algorithms of a second type of forecasting algorithm;

determining an initial confidence level and a half-life of each of the first set of forecasting algorithms and the second set of forecasting algorithms, wherein the initial confidence level describes an accuracy level of each of the forecasting algorithms at an initial time in predicting a condition, and wherein the half-life describes a subsequent time at which a forecasting algorithm has reached half of its forecast horizon;

determining a half-life weight for each of the first set of forecasting algorithms and the second set of forecasting algorithms at a subsequent time that is subsequent to the initial time, wherein half-life weights decrease an effect of a forecasting algorithm as time elapses;

determining a combined confidence level of the forecasting cohort at the subsequent time, wherein the combined confidence level is based on the initial confidence level and the half-life weight of each of the first set of forecasting algorithms and the second set of forecasting algorithms; and

utilizing the combined confidence level of the forecasting cohort at the subsequent time to adjust resource usage.

12. The computer program product of claim 11, wherein the method further comprises:

receiving sensor readings from a device, wherein the sensor readings describe current conditions of the device, and wherein the device includes resources;

predicting future usage of the resources in the device based on the combined confidence level of the forecasting cohort; and

reallocating the resources used by the device based on the predicted future usage.

13. The computer program product of claim 12, wherein the resources used by the device are software resources.

14. The computer program product of claim 12, wherein the resources used by the device are hardware resources.

15. The computer program product of claim 12, wherein reallocated resources are in the device.

16. The computer program product of claim 12, wherein the sensor readings are received at the initial time.

17. The computer program product of claim 12, wherein the sensor readings are received at the subsequent time.

18. The computer program product of claim 11, wherein the half-life weight of a forecasting algorithm is:

19. The computer program product of claim 11, wherein the resource usage is predicted at the subsequent time.

20. A device comprising:

a sensor, wherein the sensor develops sensor readings from the device that describe current conditions of the device;

one or more processors for generating a forecasting cohort of forecasting algorithms, wherein the forecasting cohort comprises a first set of forecasting algorithms of a first type of forecasting algorithm, and wherein the forecasting cohort further comprises a second set of forecasting algorithms of a second type of forecasting algorithm;

one or more processors for determining an initial confidence level and a half-life of each of the first set of forecasting algorithms and the second set of forecasting algorithms, wherein the initial confidence level describes an accuracy level of each of the forecasting algorithms at an initial time in predicting a condition, and wherein the half-life describes a subsequent time at which a forecasting algorithm has reached half of its forecast horizon;

one or more processors for determining a half-life weight for each of the first set of forecasting algorithms and the second set of forecasting algorithms at a subsequent time that is subsequent to the initial time, wherein half-life weights decrease an effect of a forecasting algorithm as time elapses;

one or more processors for determining a combined confidence level of the forecasting cohort at the subsequent time, wherein the combined confidence level is based on the initial confidence level and the half-life weight of each of the first set of forecasting algorithms and the second set of forecasting algorithms;

one or more processors for predicting future usage of the resources by the device based on the combined confidence level of the forecasting cohort; and

one or more processors for reallocating resources based on the predicted future usage, wherein the half-life weight is calculated according to;