Risk analysis for data-intensive stochastic models

• US 8,341,180 B1
• Filed: 09/13/2011
• Issued: 12/25/2012
• Est. Priority Date: 09/13/2011
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation, Rice University, University of Florida (State University System of Florida)
• Inventors: Haas, Peter J., Xu, Fei, Arumugam, Subramanian, Jampani, Ravindranath Chowdary, Jermaine, Christopher, Perez, Luis L
• Primary Examiner(s): LEWIS, CHERYL RENEA
• Application Number:
  US13/231,792
• Time in Patent Office: 469 Days
• Field of Search: 707/610, 707/770, 707/805
• US Class Current: 707/770
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202

Abstract

A risk analysis system and method are provided. The system includes an analyzer for analyzing database instances by executing a query on each database instance and selecting a cutoff value. The analyzer also discards the sets of uncertainty data that yield query-result values below the cutoff value and retains the database instances that yield query-result values above the cutoff value as elite sets. The system also includes a cloner to replicate the elite sets, and a sampler to modify the elite sets so that each elite set is mutually statistically independent while still yielding query-result values above the cutoff value.

Claims

1. A computer program product comprising:

a computer readable storage medium to store a computer readable program, wherein the computer readable program, when executed by a processor within a computer, causes the computer to perform operations for designating a graphical element displayed in a user interface, the operations comprising;

generating a plurality of database instances from a possible-worlds database according to a plurality of Monte Carlo variable generation functions;

analyzing the database instances from the possible-worlds database by executing a query on each database instance to generate a query-result distribution having an extreme upper quantile of interest;

selecting a cutoff value; and

discarding the database instances that yield query-result values below the cutoff value, and retaining the database instances that yield query-result values above the cutoff value as elite sets.

2. The computer program product of claim 1, wherein the operations further comprise cloning the elite sets.

3. The computer program product of claim 2, wherein the operations further comprise modifying the elite sets so that each elite set is mutually statistically independent while still yielding query-result values above the cutoff value.

4. The computer program product of claim 3, wherein modifying the elite sets further comprises:

selecting an element from one of the elite sets;

generating a trial element to replace the element;

analyzing the elite set to determine if the query-result value is greater than the cutoff value; and

replacing the element with the trial value if the query-result value exceeds the cutoff value.

5. The computer program product of claim 1, wherein the operations further comprise updating the cutoff value after each iteration of analyzing and discarding until the cutoff value coincides with a predefined extreme upper quantile.

6. The computer program product of claim 5, wherein all intermediate tail probabilities are equal and all intermediate sample sizes are equal.

7. A method comprising:

generating a plurality of database instances from a possible-worlds database according to a plurality of Monte Carlo variable generation functions;

analyzing the database instances from the possible-worlds database by executing a query on each database instance to generate a query-result distribution having an extreme upper quantile of interest;

selecting a cutoff value; and

discarding the database instances that yield query-result values below the cutoff value, and retaining the database instances that yield query-result values above the cutoff value as elite sets.

8. The method of claim 7, further comprising cloning the elite sets.

9. The method of claim 8, further comprising modifying the elite sets so that each elite set is mutually statistically independent while still yielding query-result values above the cutoff value.

10. The method of claim 9, wherein modifying the elite sets further comprises:

selecting an element from one of the elite sets;

generating a trial element to replace the element;

analyzing the elite set to determine if the query-result value is greater than the cutoff value; and

replacing the element with the trial element if the query-result value is greater than.

11. The method of claim 7, further comprising updating the cutoff value after each iteration of analyzing and discarding until the cutoff value coincides with a predefined extreme upper quantile.

12. The method of claim 11, wherein the cutoff value is equal to a 50% quantile of query-result values of the elite sets.

13. A system comprising:

a Monte Carlo database (MCDB) system use a variable generation (VG) function to generate a plurality of database instances from a possible-worlds database according to a plurality of Monte Carlo variable generation functions; and

an analyzer configured to analyze the database instances from the possible-worlds database by executing a query on each database instance to generate a query-result distribution having an extreme quantile of interest;

wherein the analyzer is further configured to;

select a cutoff value;

discard the database instances that yield query-result values below the cutoff value; and

retain the database instances that yield query-result values above the cutoff value as elite sets.

14. The system of claim 13, further comprising a cloner configured to replicate the elite sets.

15. The system of claim 14, further comprising a sampler configured to modify the elite sets so that each elite set is mutually statistically independent while still yielding query-result values above the cutoff value.

16. The system of claim 15, wherein the sampler is further configured to:

select an element from one of the elite sets;

generate a trial element to replace the element;

analyze the elite set to determine if the query-result value is greater than the cutoff value; and

replace the element with the trial element if the query-result value is greater than the cutoff value.

17. The system of claim 13, further comprising a rare event generator configured to update the cutoff value after each iteration of analyzing and discarding until a desired quantile portion is reached.

18. The system of claim 17, wherein all intermediate tail probabilities are equal and all intermediate sample sizes are equal.