PERFORMANCE TESTING BASED ON VARIABLE LENGTH SEGMENTATION AND CLUSTERING OF TIME SERIES DATA

• US 20170147930A1
• Filed: 01/30/2015
• Published: 05/25/2017
• Est. Priority Date: 01/30/2015
• Status: Active Grant

• Current Assignee: Micro Focus LLC (Open Text Corporation)
• Original Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Inventors: BELLALA, Gowtham, ZHANG, Mi, LYON, Geoff M.
• Granted Patent: US 10,643,138 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

Performance testing based on variable length segmentation and clustering of time series data is disclosed. One example is a system including a training module, a performance testing module, and an interface module. The training module generates a trained model to learn characteristics of a system of interest from training time series data by segmenting the training time series data into homogeneous windows of variable length, clustering the segments to identify patterns, and associating each cluster with a cluster score. The performance testing module analyzes system characteristics from testing time series data by receiving the testing time series data, and determining a performance metric for the testing time series data by analyzing the testing time series data based on the trained model. The interface module is communicatively linked to the performance testing module, and provides the performance metric via an interactive graphical user interface.

Claims

1. A system comprising:

a training module to generate a trained model to learn characteristics of a system of interest from training time series data by;

segmenting the training time series data into homogeneous windows of variable length,

clustering the segments to identify patterns, and

associating each cluster with a cluster score;

a performance testing module to analyze system characteristics from testing time series data by;

receiving the testing time series data, and

determining a performance metric for the testing time series data by analyzing the testing time series data based on the trained model; and

an interface module, communicatively linked to the performance testing module, to provide the performance metric via an interactive graphical user interface.

2. The system of claim 1, wherein clustering the segments is based on one of segment categorization and feature space transformation.

3. The system of claim 1, wherein the interface module further receives the cluster score from a domain expert via the interactive graphical user interface or automatically generates the cluster score based on predetermined rules.

4. The system of claim 1, wherein the training module further pre-processes the testing time series data based on one of data interpolation, data imputation, feature sub-selection, data fusion, validation, estimation, and editing.

5. The system of claim 1, wherein the performance testing module further:

generates additional segments of the testing time series data;

associates each additional segment with a training cluster in the trained model; and

associates each additional segment with a segment score based on the cluster score for the associated training cluster.

6. The system of claim 5, wherein the performance metric is based on a weighted aggregate of the segment scores.

7. The system of claim 1, wherein the characteristics of the system of interest include one of operational and behavioral performance of the system.

8. The system of claim 7, wherein the performance metric is a behavior scorecard for a vehicle driver.

9. The system of claim 8, wherein associating each cluster with a cluster score is based on a variance of a load value.

10. A method for determining a performance metric for a system, the method comprising:

pre-processing, via a processor, testing time series data related to a vehicular driving record under performance testing;

segmenting the testing time series data based on windows of variable length to determine homogeneous time windows;

associating each segment with a training cluster in a trained model, the trained model generated based on training time series data;

determining a segment score for each segment based on a cluster score for the associated training cluster;

determining a performance metric for the testing time series data, the performance metric based on a weighted aggregate of the segment scores, and the performance metric indicative of a quality of driving; and

providing the performance metric via an interactive graphical user interface.

11. The method of claim 10, wherein associating each segment with the training cluster is based on a Euclidean distance of the segment to the training cluster in a feature space comprising labeled training data from the trained model.

12. The method of claim 10, wherein the training clusters are based on durational partitions of the training time series data, and wherein associating each segment with the training cluster further comprises identifying a durational partition for the segment.

13. The method of claim 10, further comprising automatic generation of a cluster score for a given cluster based on an identified pattern for the given cluster.

14. The method of claim 10, wherein pre-processing the testing time series data further includes one of data interpolation, data imputation, feature sub-selection, data fusion, validation, estimation, and editing.

15. A non-transitory computer readable medium comprising executable instructions to:

pre-process training time series data received from a plurality of sources;

generate a trained model based on the pre-processed training time series data by;

segmenting, via a processing system, the pre-processed training time series data into homogeneous windows of variable length, clustering the segments to identify patterns, and

associating each cluster with a cluster score;

receive testing time series data;

determine a performance metric for testing time series data by analyzing the testing time series data based on the trained model; and

provide the performance metric via an interactive graphical user interface.