Automated time series forecasting pipeline generation

Patent metadata

Field	Value
Publication number	US-11966340-B2
Application number	US-202217654965-A
Country	US
Kind code	B2
Filing date	Mar 15, 2022
Priority date	Feb 18, 2021
Publication date	Apr 23, 2024
Grant date	Apr 23, 2024

Abstract

To automate time series forecasting machine learning pipeline generation, a data allocation size of time series data may be determined based on one or more characteristics of a time series data set. The time series data may be allocated for use by candidate machine learning pipelines based on the data allocation size. Features for the time series data may be determined and cached by the candidate machine learning pipelines. Predictions of each of the candidate machine learning pipelines using at least the one or more features may be evaluated. A ranked list of machine learning pipelines may be automatically generated from the candidate machine learning pipelines for time series forecasting based upon evaluating predictions of each of the one or more candidate machine learning pipelines.

First claim

What is claimed is:

1. A method for automated time series forecasting machine learning pipeline generation in a computing environment by one or more processors comprising:
determining a data allocation size of time series data based on one or more characteristics of a time series data set;
executing machine learning logic to allocate the time series data to one or more candidate machine learning pipelines based on the data allocation size, wherein a sequential order of the time series data set is used while allocating the time series data based on the data allocation size;
executing the machine learning logic to train the one or more candidate machine learning pipelines using selected data associated with the time series data set;
executing the machine learning logic to determine and cache one or more features for the time series data by the one or more candidate machine learning pipelines;
evaluating predictions of each of the one or more candidate machine learning pipelines using at least the one or more features; and
automatically generating a ranked list of machine learning pipelines from the one or more candidate machine learning pipelines for time series forecasting based upon evaluating predictions of each of the one or more candidate machine learning pipelines.

2. The method of claim 1 , further including determining a holdout data set, a test data set, and a training data set from the time series data for allocating the time series data.

3. The method of claim 1 , further including evaluating the trained one or more candidate machine learning pipelines using the time series data, a hold data set, a test data set, and a training data set from the time series data.

4. The method of claim 1 , further including allocating the time series data backward in time.

5. The method of claim 1 , further including combining the one or more features with previously determined features for use by the one or more candidate machine learning pipelines.

6. The method of claim 1 , further including caching the one or more features at a final estimator of the one or more candidate machine learning pipelines.

7. A system for automated time series forecasting machine learning pipeline generation in a computing environment, comprising:
one or more computers with executable instructions that when executed cause the system to:
determine a data allocation size of time series data based on one or more characteristics of a time series data set;
execute machine learning logic to allocate the time series data to one or more candidate machine learning pipelines based on the data allocation size, wherein a sequential order of the time series data set is used while allocating the time series data based on the data allocation size;
execute the machine learning logic to train the one or more candidate machine learning pipelines using selected data associated with the time series data set;
execute the machine learning logic to determine and cache one or more features for the time series data by the one or more candidate machine learning pipelines;
evaluate predictions of each of the one or more candidate machine learning pipelines using at least the one or more features; and
automatically generate a ranked list of machine learning pipelines from the one or more candidate machine learning pipelines for time series forecasting based upon evaluating predictions of each of the one or more candidate machine learning pipelines.

8. The system of claim 7 , wherein the executable instructions when executed cause the system to determine a holdout data set, a test data set, and a training data set from the time series data for allocating the time series data.

9. The system of claim 7 , wherein the executable instructions when executed cause the system to evaluate the trained one or more candidate machine learning pipelines using the time series data, a hold data set, a test data set, and a training data set from the time series data.

10. The system of claim 7 , wherein the executable instructions when executed cause the system to allocate the time series data backward in time.

11. The system of claim 7 , wherein the executable instructions when executed cause the system to combine the one or more features with previously determined features for use by the one or more candidate machine learning pipelines.

12. The system of claim 7 , wherein the executable instructions when executed cause the system to cache the one or more features at a final estimator of the one or more candidate machine learning pipelines.

13. A computer program product for automated time series forecasting machine learning pipeline generation in a computing environment, the computer program product comprising:
one or more computer non-transitory readable storage media, and program instructions collectively stored on the one or more non-transitory computer readable storage media, the program instructions comprising:
program instructions to determine a data allocation size of time series data based on one or more characteristics of a time series data set;
program instructions to execute machine learning logic to allocate the time series data to one or more candidate machine learning pipelines based on the data allocation size, wherein a sequential order of the time series data set is used while allocating the time series data based on the data allocation size;
program instructions to execute the machine learning logic to train the one or more candidate machine learning pipelines using selected data associated with the time series data set;
program instructions to execute the machine learning logic to determine and cache one or more features for the time series data by the one or more candidate machine learning pipelines;
program instructions to evaluate predictions of each of the one or more candidate machine learning pipelines using at least the one or more features; and
program instructions to automatically generate a ranked list of machine learning pipelines from the one or more candidate machine learning pipelines for time series forecasting based upon evaluating predictions of each of the one or more candidate machine learning pipelines.

14. The computer program product of claim 13 , further including program instructions to:
determine a holdout data set, a test data set, and a training data set from the time series data for allocating the time series data; and
evaluate the trained one or more candidate machine learning pipelines using the time series data, the hold data set, the test data set, and the training data set from the time series data.

15. The computer program product of claim 13 , further including program instructions to allocate the time series data backward in time.

16. The computer program product of claim 13 , further including program instructions to combine the one or more features with previously determined features for use by the one or more candidate machine learning pipelines.

17. The computer program product of claim 13 , further including program instructions to cache the one or more features at a final estimator of the one or more candidate machine learning pipelines.

Assignees

• IBM

Inventors

• Vu Long
• Chen Bei
• Dang Xuan-Hong
• Kirchner Peter Daniel
• Shah Syed Yousaf
• Patel Dhavalkumar C
• Han Si Er
• Yang ji hui
• Wang Jun
• Xu Jing James
• Wang Dakuo
• Bramble Gregory
• Samulowitz Horst Cornelius
• Sathe Saket K
• Gifford Wesley M
• Zerfos Petros

Classifications

• G06F12/0871 Primary

  Allocation or management of cache space · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F2212/604

  Details relating to cache allocation · CPC title

• G06N20/10 Primary

  using kernel methods, e.g. support vector machines [SVM] · CPC title

• G06F12/0862

  with prefetch · CPC title