Time-series data analyzing apparatus

• US 7,437,266 B2
• Filed: 03/28/2006
• Issued: 10/14/2008
• Est. Priority Date: 08/31/2005
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation), Toshiba Digital Solutions Corporation (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Orihara, Ryohei, Ueno, Ken, Kitahara, Youichi
• Primary Examiner(s): Bui; Bryan
• Application Number:
  US11/390,164
• Publication Number:
  US 20070079195A1
• Time in Patent Office: 931 Days
• Field of Search: 702/176, 702/179, 702/181, 702/182, 702/183, 702/187, 707/6, 707/7, 700/90, 714/731
• US Class Current: 702/176
• CPC Class Codes: A61B 5/021 Measuring pressure in heart or blood vessels A61B5/0205 takes precedence
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/00 Aspects of pattern recognition specially adapted for signal processing

Abstract

A time-series data analyzing apparatus which extracts a composite factor time-series pattern from time-series data. The apparatus includes a dividing device which divides the time-series data into pattern generation time-series data and pattern inspection time-series data which do not include pattern generation time-series data. A first generating device generates a transitional pattern including a support time data indicating a transition of support time and having a transition occurrence probability higher than a minimum occurrence probability in the pattern generation time-series data. A second generating device generates frequently appearing integrated transitional patterns. A second computing device computes cause-and-effect strength of each of the frequently appearing integrated transitional patterns using the pattern inspection time-series data. A display device displays the composite factor time-series pattern having the cause-and-effect strength higher than the minimum cause-and-effect strength given preliminarily.

Claims

1. A time-series data analyzing apparatus which extracts a composite factor time-series pattern indicating a composite factor of a focused transitional pattern from time-series data, the apparatus comprising:

a dividing device configured to divide the time-series data to pattern generation time-series data and pattern inspection time-series data which do not include pattern generation time-series data;

a first generating device configured to generate a transitional pattern including a support time data indicating a transition of support time and having a transition occurrence probability higher than a minimum occurrence probability in the pattern generation time-series data;

a first computing device configured to compute a time lag range specified by a maximum value and a minimum value of support time in the support time data;

a second generating device configured to generate a plurality of frequently appearing integrated transitional patterns by executing integration based on pattern matching between a plurality of transitional patterns obtained by adding a time lag value selected within the time lag range to support time data of the transitional pattern and the focused transitional pattern;

a second computing device configured to compute cause-and-effect strength of each of the plurality of frequently appearing integrated transitional patterns using the pattern inspection time-series data; and

a display device to display the composite factor time-series pattern having the cause-and-effect strength higher than the minimum cause-and-effect strength given preliminarily.

2. The time-series data analyzing apparatus according to claim 1, further comprising a discretizing device configured to discretize multivariate time-series data based on a preliminarily provided standard, the multivariate time-series data being included in the time-series data.

3. The time-series data analyzing apparatus according to claim 1, further comprising:

a second generating device configured to generate time-series event data from the pattern generation time-series data; and

an extracting device configured to extract a frequently appearing single item having an item occurrence probability higher than the minimum occurrence probability in the pattern generation time-series data from the time-series event data,

wherein the first generating device is configured to generate the transitional pattern based on a plurality of transitional pattern candidates obtained by self-joining of the frequently appearing single items.

4. The time-series data analyzing apparatus according to claim 1, further comprising:

an operation device to receive an instruction from a user, wherein the display device rearranges and displays the plurality of composite factor time-series patterns corresponding to the instruction.

5. The time-series data analyzing apparatus according to claim 1, further comprising:

an inspecting device configured to inspect whether each of the plurality of frequently appearing integrated transitional patterns satisfies time cause-and-effect property.

6. The time-series data analyzing apparatus according to claim 1, wherein the second computing device is configured to disassemble one of the plurality of frequently appearing integrated transitional patterns to a cause preceding pattern and a result receding pattern;

compute a cause preceding occurrence probability of the cause preceding pattern and an effect preceding occurrence probability of the result preceding pattern from the pattern inspection time-series data; and

wherein said cause-and-effect strength includes a ratio between the cause preceding occurrence probability and the effect preceding occurrence probability.

7. The time-series data analyzing apparatus according to claim 1, wherein the second computing device is configured to pick up only data in which a difference is recognized using statistical inspection of discrete quantitative value as said composite factor time-series pattern.

8. The time-series data analyzing apparatus according to claim 1, wherein the second computing device is configured to pick up only data in which a difference in symmetry property of transition is recognized as said composite factor time-series pattern.

9. A time-series data analyzing method for extracting a composite factor time-series pattern indicating a composite factor of a focused transitional pattern from time-series data, comprising:

dividing the time-series data to pattern generation time-series data and pattern inspection time-series data which do not include pattern generation time-series data;

generating a transitional pattern including a support time data indicating a transition of support time and having a transition occurrence probability higher than a minimum occurrence probability in the pattern generation time-series data;

computing a time lag range specified by a maximum value and a minimum value of support time in the support time data;

generating plurality of frequently appearing integrated transitional patterns by executing integration based on pattern matching between a plurality of transitional patterns obtained by adding a time lag value selected within the time lag range to support time data of the transitional pattern and the focused transitional pattern;

computing cause-and-effect strength of each of the plurality of frequently appearing integrated transitional patterns using the pattern inspection time-series data; and

displaying the composite factor time-series pattern having the cause-and-effect strength higher than the minimum cause-and-effect strength given preliminarily.

10. The time-series data analyzing method according to claim 9, further comprising discretizing multivariate time-series data based on a preliminarily provided standard, the multivariate time-series data being included in the time-series data.

11. The time-series data analyzing method according to claim 9, further comprising:

generating time-series event data from the pattern generation time-series data; and

extracting a frequently appearing single item having an item occurrence probability higher than the minimum occurrence probability in the pattern generation time-series data from the time-series event data, wherein the transitional pattern is generated based on a plurality of pattern candidates obtained by self-joining of the frequently appearing single items.

12. The time-series data analyzing method according to claim 9, further comprising:

receiving an instruction from a user, wherein the displaying includes rearranging and displaying the plurality of composite factor time-series patterns corresponding to the instruction.

13. The time-series data analyzing method according to claim 9, further comprising:

inspecting whether each of the plurality of frequently appearing integrated transitional patterns satisfies time cause-and-effect property.

14. The time-series data analyzing method according to claim 9, further comprising:

disassembling one of the plurality of frequently appearing integrated transitional patterns to a cause preceding pattern and a result preceding pattern; and

computing a cause preceding occurrence probability of the cause preceding pattern and an effect preceding occurrence probability of the result preceding pattern from the pattern inspection time-series data, wherein said cause-and-effect strength includes a ratio between the cause preceding occurrence probability and the effect preceding occurrence probability.

15. The time-series data analyzing method according to claim 9, wherein said composite factor time-series pattern includes data in which a difference is recognized using statistical inspection of discrete quantitative value.

16. The time-series data analyzing method according to claim 9, wherein said composite factor time-series pattern includes data in which a difference in symmetry property of transition is recognized.