Comparing data series associated with two systems to identify hidden similarities between them

• US 8,738,563 B2
• Filed: 03/28/2010
• Issued: 05/27/2014
• Est. Priority Date: 03/28/2010
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Shehory, Onn Menahem, Shochat, Moran, Farchi, Eitan Daniel
• Primary Examiner(s): Gaffin, Jeffrey A
• Assistant Examiner(s): Sugent, James F
• Application Number:
  US12/748,415
• Publication Number:
  US 20110238613A1
• Time in Patent Office: 1,521 Days
• Field of Search: 706/52, 702/186
• US Class Current: 706/52
• CPC Class Codes: G06F 11/3409 for performance assessment
• CPC Class Codes: G06F 11/3419 by assessing time
• CPC Class Codes: G06F 11/3664 Environments for testing or debugging software
• CPC Class Codes: G06F 11/3676 for coverage analysis
• CPC Class Codes: G06F 11/3696 Methods or tools to render software testable
• CPC Class Codes: G06F 17/15 Correlation function computation including computation of convolution operations arithmetic circuits for sum of products per se, e.g. multiply-accumulators G06F7/5443; digital filters, e.g. FIR, IIR, adaptive filters H03H17/00
• CPC Class Codes: G06N 5/048 Fuzzy inferencing
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A method that includes: sampling a plurality of parameters associated with a first and a second system to yield a first and a second set of data series, respectively, each data series being associated with samples of a respective parameter; applying, for each two respective data series, a transformation associated with a specified order selected from a plurality of ordered transformations, such that the selected transformation is applied repeatedly to at least one of the two respective data series, wherein, in each repeated application, the transformation is assigned with a different value of a transformation parameter; determining, for each parameter, a one of the plurality of transformation parameters for which the correlation level of the two respective time series is above a specified value; and repeating the applying and the determining with a transformation that is associated with a higher order of the ordered transformations.

Claims

1. A method comprising:

obtaining a first data series and a second data series, wherein the first data series being associated with values of a parameter sampled from a first system, wherein the second data series being associated with values of the parameter sampled from a second system;

calculating a correlation level between the first and second data series;

in response to the correlation level being below a predetermined threshold, processing the first and second data series by;

repeatedly selecting a transformation from a transformation repository, applying the transformation to the second data series, and computing the correlation level of the first data series and the transformed second data series;

selecting, based on the computed correlation levels, the transformation of the plurality of transformations that yields a correlation level above the predetermined threshold; and

providing a recommendation of a modification to the second system based on the selected transformation;

wherein the plurality of transformations are associated with an order, and wherein said selecting the transformation from the plurality of transformations is performed based on the order, and

reordering the transformations in the transformation repository to reduce computational intensity of the repeated applications;

wherein at least one of said calculating, said applying and said providing is performed by a processor.

2. The method of claim 1 further comprising applying the recommendation on the second system, thereby increasing similarity between the first and second systems.

3. The method of claim 1, wherein at least two transformations in the transformation repository differ by a transformation parameter.

4. The method of claim 1 further comprises updating the transformations in the repository in view of a transformation parameter of the selected transformation.

5. The method of claim 1, wherein the order is modified based on user selections over time based on automatic learning.

6. The method of claim 1 further comprises partitioning the first and second data series, wherein the correlation of a first corresponding pair of segments of the first and second data series having a correlation above the predetermined threshold, wherein the correlation of a second corresponding pair of segments of the first and second data series having a correlation below the predetermined threshold, and wherein said processing is performed with respect to the second corresponding pair of segments.

7. The method of claim 6, wherein the recommendation is associated with a portion of a time of execution of the second system.

8. The method of claim 1, wherein the transformations of the repository are configured to simulate configuration change of the second system.

9. The method of claim 1, wherein the first system is a computer system and the second system is a test system.

10. The method of claim 1, wherein the transformations are configured to yield a similarity analysis of respective parameters between the test and the computer systems.

11. The method of claim 1, wherein the correlation level is determined by computing a Pearson product moment of the first and second data series.

12. The method of claim 1, wherein at least some of the transformations comprise at least one of:

linear series manipulations;

non-linear series manipulations;

mapping of at least one of the series from time domain into frequency domain.

13. The method of claim 1, wherein the second data series comprises data points assigned along a unit axis, and wherein at least one of the transformations comprises multiplying the unit axis of the second data series with a specified value.

14. The method of claim 13, wherein the unit axis is associated with time.

15. The method of claim 1, wherein the first and second data series are time series associated with a specified time base.

16. The method of claim 1 further comprising monitoring the first and second systems to sample values of the parameter during execution thereof, wherein said obtaining comprises obtaining the monitored values.

17. A computer program product comprising a non-transitory computer readable medium having instructions retained thereon, which instructions, when provided to a processor, cause the processor to:

obtain a first data series and a second data series, wherein the first data series being associated with values of a parameter sampled from a first system, wherein the second data series being associated with values of the parameter sampled from a second system;

calculate a correlation level between the first and second data series;

in response to the correlation level being below a predetermined threshold, process the first and second data series by;

repeatedly selecting a transformation from a transformation repository, applying the transformation to the second data series, and computing the correlation level of the first data series and the transformed second data series;

selecting, based on the computed correlation levels, the transformation of the plurality of transformations that yields a correlation level above the predetermined threshold; and

providing a recommendation of a modification to the second system based on the selected transformation;

wherein the plurality of transformations are associated with an order, and wherein said selecting the transformation from the plurality of transformations is performed based on the order, and

reordering the transformations in the transformation repository to reduce computational intensity of the repeated applications.

18. A computerized apparatus comprising a processor coupled to a memory, wherein said processor is configured to:

obtain a first data series and a second data series, wherein the first data series being associated with values of a parameter sampled from a first system, wherein the second data series being associated with values of the parameter sampled from a second system;

calculate a correlation level between the first and second data series;

in response to the correlation level being below a predetermined threshold, process the first and second data series by;

repeatedly selecting a transformation from a transformation repository, applying the transformation to the second data series, and computing the correlation level of the first data series and the transformed second data series;

selecting, based on the computed correlation levels, the transformation of the plurality of transformations that yields a correlation level above the predetermined threshold; and

providing a recommendation of a modification to the second system based on the selected transformation;

wherein the plurality of transformations are associated with an order, and wherein said selecting the transformation from the plurality of transformations is performed based on the order, and reordering the transformations in the transformation repository to reduce computational intensity of the repeated applications.