Semi-automatic system with an iterative learning method for uncovering the leading indicators in business processes

• US 8,010,589 B2
• Filed: 02/20/2007
• Issued: 08/30/2011
• Est. Priority Date: 02/20/2007
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: Peng, Wei, Rose, Philip C., Sun, Tong
• Primary Examiner(s): Bullock, Jr., Lewis
• Assistant Examiner(s): SANDIFER, MATTHEW D
• Application Number:
  US11/676,816
• Publication Number:
  US 20080201397A1
• Time in Patent Office: 1,652 Days
• Field of Search: 708422-426, 705/7.39
• US Class Current: 708/422
• CPC Class Codes: G06Q 10/00 Administration; Management
• CPC Class Codes: G06Q 10/0637 Strategic management or analysis, e.g. setting a goal or target of an organisation; Planning actions based on goals; Analysis or evaluation of effectiveness of goals
• CPC Class Codes: G06Q 10/0639 Performance analysis of employees; Performance analysis of enterprise or organisation operations
• CPC Class Codes: G06Q 10/06393 Score-carding, benchmarking or key performance indicator [KPI] analysis

Abstract

Embodiments herein select performance indicators from raw data and measure the indicators over at least one time period to extract a time series of data for each of the indicators. The method determines a time order among the correlated indicators and determines a causal direction among the correlated indicators based on which of the correlated indicators occurs first in time so as to identify relative leading indicators among the correlated indicators. The processes of determining the time order and determining the causal direction can comprise applying Dynamic Time Warping and/or Granger Causality techniques to the time series of data.

Claims

1. A computer-implemented method comprising:

defining, by a computer, data points from a model;

producing raw data of production operations corresponding to said data points, said raw data being produced by production machines used in said production operations;

computing, by said computer, performance indicators from said raw data;

measuring, by said computer, said indicators over at least one time period to extract a time series of data for each of said indicators;

filtering out, by said computer, redundant indicators to produce a reduced indicator set of time series of data;

detecting, by said computer, correlations among said time series of data within said reduced indicator set by considering time-shifts between said time series of data so as to identify correlated indicators;

determining, by said computer, a time order among said correlated indicators;

determining, by said computer, a causal direction among said correlated indicators so as to identify relative leading indicators among said correlated indicators;

creating, by said computer, a similarity matrix among said correlated indicators based on said time order and said causal direction among said correlated indicators;

partitioning, by said computer, said correlated indicators within said similarity matrix into clusters using an agglomerative clustering process;

identifying, by said computer, said relative leading indicators within each cluster as root leading indicators of a each of said clusters; and

producing, by said computer, a report of said root leading indicators of said production operations.

2. The method according to claim 1, wherein said filtering comprises applying at least one unsupervised dimensionality technique to said times series of data.

3. The method according to claim 1, wherein said filtering comprises applying at least one Principle Component Analysis and Singular Value Decomposition techniques to said times series of data.

4. The method according to claim 1, wherein said determining of said time order and said determining of said causal direction comprise applying a Dynamic Time Warping technique to said time series of data.

5. The method according to claim 1, wherein said determining of said time order and said determining of said causal direction comprise applying Dynamic Time Warping and Granger Causality techniques to said time series of data.

6. The method according to claim 1, wherein said determining of said causal direction comprises applying a Granger Causality technique to said time series of data.

7. A computer-implemented method comprising:

defining, by a computer, data points from a model;

producing raw data of production operations corresponding to said data points, said raw data being produced by production machines used in said production operations;

computing, by said computer, performance indicators from said raw data;

measuring, by said computer, said indicators over at least one time period to extract a time series of data for each of said indicators;

filtering out, by said computer, redundant indicators to produce a reduced indicator set of time series of data;

detecting, by said computer, correlations among said time series of data within said reduced indicator set by considering time-shifts between said time series of data so as to identify correlated indicators;

determining, by said computer, a time order among said correlated indicators;

determining, by said computer, a causal direction among said correlated indicators based on which of said correlated indicators occurs first in time so as to identify relative leading indicators among said correlated indicators;

creating, by said computer, a similarity matrix among said correlated indicators based on said time order and said causal direction among said correlated indicators;

partitioning, by said computer, said correlated indicators within said similarity matrix into clusters using an agglomerative clustering process;

identifying, by said computer, said relative leading indicators within each cluster as root leading indicators of a each of said clusters; and

producing, by said computer, a report of said root leading indicators of said production operations.

8. The method according to claim 7, wherein said filtering comprises applying at least one unsupervised dimensionality technique to said times series of data.

9. The method according to claim 7, wherein said filtering comprises applying at least one Principle Component Analysis and Singular Value Decomposition techniques to said times series of data.

10. The method according to claim 7, wherein said determining of said time order and said determining of said causal direction comprise applying a Dynamic Time Warping technique to said time series of data.

11. The method according to claim 7, wherein said determining of said time order and said determining of said causal direction comprise applying Dynamic Time Warping and Granger Causality techniques to said time series of data.

12. The method according to claim 7, wherein said determining of said causal direction comprises applying a Granger Causality technique to said time series of data.

13. A computer-implemented method comprising:

defining, by a computer, data points from a model;

producing raw data of printing operations corresponding to said data points, said raw data being produced by printing machines used in said printing operations;

computing, by said computer, performance indicators from said raw data;

measuring, by said computer, said indicators over at least one time period to extract a time series of data for each of said indicators;

filtering out, by said computer, redundant indicators to produce a reduced indicator set of time series of data;

detecting, by said computer, correlations among said time series of data within said reduced indicator set by considering time-shifts between said time series of data so as to identify correlated indicators;

determining, by said computer, a time order among said correlated indicators;

determining, by said computer, a causal direction among said correlated indicators based on a relative ability of each of said indicators to predict behavior of another of said correlated indicators so as to identify relative leading indicators among said correlated indicators;

creating, by said computer, a similarity matrix among said correlated indicators based on said time order and said causal direction among said correlated indicators;

partitioning, by said computer, said correlated indicators within said similarity matrix into clusters using an agglomerative clustering process;

identifying, by said computer, said relative leading indicators within each cluster as root leading indicators of a each of said clusters; and

producing, by said computer, a report of said root leading indicators of said printing operations.

14. The method according to claim 13, wherein said filtering comprises applying at least one unsupervised dimensionality technique to said times series of data.

15. The method according to claim 13, wherein said filtering comprises applying at least one Principle Component Analysis and Singular Value Decomposition techniques to said times series of data.

16. The method according to claim 13, wherein said determining of said time order and said determining of said causal direction comprise applying a Dynamic Time Warping technique to said time series of data.

17. The method according to claim 13, wherein said determining of said time order and said determining of said causal direction comprise applying Dynamic Time Warping and Granger Causality techniques to said time series of data.

18. The method according to claim 13, wherein said determining of said causal direction comprises applying a Granger Causality technique to said time series of data.

19. A computer-implemented method comprising:

defining, by a computer, data points from a model;

producing raw data of production operations corresponding to said data points, said raw data being produced by production machines used in said production operations;

computing, by said computer, performance indicators from said raw data;

measuring, by said computer, said indicators over at least one time period to extract a time series of data for each of said indicators;

filtering out, by said computer, redundant indicators to produce a reduced indicator set of time series of data;

detecting, by said computer, correlations among said time series of data within said reduced indicator set by considering time-shifts between said time series of data so as to identify correlated indicators;

determining, by said computer, a time order among said correlated indicators;

determining, by said computer, a causal direction among said correlated indicators based on which of said correlated indicators occurs first in time so as to identify relative leading indicators among said correlated indicators, wherein if said correlated indicators occur at approximately a same time, said determining of said causal direction is based on a relative ability of each of said indicators to predict behavior of another of said correlated indicators;

creating, by said computer, a similarity matrix among said correlated indicators based on said time order and said causal direction among said correlated indicators;

partitioning, by said computer, said correlated indicators within said similarity matrix into clusters using an agglomerative clustering process;

identifying, by said computer, said relative leading indicators within each cluster as root leading indicators of a each of said clusters; and

producing, by said computer, a report of said root leading indicators of said production operations.

20. A non-transitory computer readable storage medium storing instructions that, when executed by a computer, cause the computer to perform a method comprising:

defining, by a computer, data points from a model;

producing raw data of production operations corresponding to said data points, said raw data being produced by production machines used in said production operations;

computing performance indicators from said raw data;

measuring said indicators over at least one time period to extract a time series of data for each of said indicators;

filtering out redundant indicators to produce a reduced indicator set of time series of data;

detecting correlations among said time series of data within said reduced indicator set by considering time-shifts between said time series of data so as to identify correlated indicators;

determining a time order among said correlated indicators;

determining a causal direction among said correlated indicators so as to identify relative leading indicators among said correlated indicators;

creating a similarity matrix among said correlated indicators based on said time order and said causal direction among said correlated indicators;

partitioning said correlated indicators within said similarity matrix into clusters using an agglomerative clustering process;

identifying said relative leading indicators within each cluster as root leading indicators of a each of said clusters; and

producing a report of said root leading indicators of said production operations.