Predicting apparatus, predicting method, and computer product

• US 20070038587A1
• Filed: 06/06/2006
• Published: 02/15/2007
• Est. Priority Date: 07/27/2005
• Status: Active Grant

• Current Assignee: Fujitsu Limited
• Original Assignee: Fujitsu Limited
• Inventors: Watanabe, Masaru, Hanaoka, Hideki
• Granted Patent: US 7,668,826 B2
• Time in Patent Office: Days
• US Class Current: 706/21
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation
• CPC Class Codes: Y10S 707/941 Human sciences
• CPC Class Codes: Y10S 707/99936 Pattern matching access

Abstract

A structuring unit generates a combination pattern of the attribute values by combining an attribute value of the assessment target data with an attribute value indicating a unspecified value. A reconstructing unit searches data matching the combination pattern from the accumulation data, and generates subset data. An index setting unit sets an index indicating significance of relationship between the generated subset data with assessment data. An assessing unit selects subset data that becomes a basis of assessment, based on the set index, and performs an assessment on a prediction result based on the selected subset data.

Claims

1. A computer-readable recording medium that stores a computer program for predicting, based on accumulation data in which a correlation between a combination of attribute values and a result of the combination is accumulated, a result,of assessment target data including a new combination of the attribute values, wherein

the computer program causes a computer to execute:

structuring including

combining an attribute value of the assessment target data with an attribute value indicating a unspecified value; and

generating a combination pattern of the attribute values;

reconstructing including

searching data matching the combination pattern from the accumulation data; and

generating subset data;

setting an index indicating significance of relationship between the generated subset data with assessment data; and

assessing including

selecting subset data that becomes a basis of assessment, based on the set index; and

performing an assessment on a prediction result based on the selected subset data.

2. The computer-readable recording medium according to claim 1, wherein

the setting includes setting the index in such a manner that subset data in which a combination of attribute values is statistically valid has precedence over other subset data.

3. The computer-readable recording medium according to claim 1, wherein

the setting includes setting the index in such a manner that subset data having larger number of attribute values matching the assessment data has precedence over other subset data.

4. The computer-readable recording medium according to claim 1, wherein

the setting includes setting the index in such a manner that, when first subset data is included in second subset data, the first subset data has precedence over the second subset data.

5. The computer-readable recording medium according to claim 2, wherein

the assessing includes

preparing a plurality of methods for assessing the prediction result based on the subset data;

and

performing the assessment on the prediction result by changing the methods.

6. The computer-readable recording medium according to claim 3, wherein

the assessing includes

preparing a plurality of methods for assessing the prediction result based on the subset data;

and

performing the assessment on the prediction result by changing the methods.

7. The computer-readable recording medium according to claim 4, wherein

the assessing includes

preparing a plurality of methods for assessing the prediction result based on the subset data;

and

performing the assessment on the prediction result by changing the methods.

8. The computer-readable recording medium according to claim 1, wherein

the setting includes

preparing a plurality of methods for determining a statistic validity the combination;

and

setting the index by changing the methods.

9. The computer-readable recording medium according to claim 1, wherein

the computer program further causes the computer to execute:

simulating including repeatedly executing other procedure by changing a parameter until a predetermined condition is satisfied.

10. The computer-readable recording medium according to claim 1, further comprising:

employing a user interface capable of displaying information used for each process.

11. An apparatus for predicting, based on accumulation data in which a correlation between a combination of attribute values and a result of the combination is accumulated, a result of assessment target data including a new combination of the attribute values, the apparatus comprising:

a structuring unit that generates a combination pattern of the attribute values by combining an attribute value of the assessment target data with an attribute value indicating a unspecified value;

a reconstructing unit that searches for data matching the combination pattern from the accumulation data, and generates subset data;

an index setting unit that sets an index indicating significance of relationship between the generated subset data with assessment data; and

an assessing unit that selects subset data that becomes a basis of assessment, based on the set index, and performs an assessment on a prediction result based on the selected subset data.

12. A method of predicting, based on accumulation data in which a correlation between a combination of attribute values and a result of the combination is accumulated, a result of assessment target data including a new combination of the attribute values, the method comprising:

structuring including

combining an attribute value of the assessment target data with an attribute value indicating a unspecified value; and

generating a combination pattern of the attribute values;

reconstructing including

searching data matching the combination pattern from the accumulation data; and

generating subset data;

setting an index indicating significance of relationship between the generated subset data with assessment data; and

assessing including

selecting subset data that becomes a basis of assessment, based on the set index; and

performing an assessment on a prediction result based on the selected subset data.