Optimized parametric modeling system and method

• US 7,136,788 B2
• Filed: 10/15/2002
• Issued: 11/14/2006
• Est. Priority Date: 10/15/2002
• Status: Expired due to Fees

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: Tupper, Joseph L.
• Primary Examiner(s): PALADINI, ALBERT WILLIAM
• Application Number:
  US10/271,965
• Publication Number:
  US 20040073475A1
• Time in Patent Office: 1,491 Days
• Field of Search: 703/2, 706/21, 707/101, 700/29
• US Class Current: 703/2
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling

Abstract

A system for enabling optimization of a parametric modeling process. The system includes a processor and an interface that allows at least one user input. Additionally, the system includes a regression analysis tree program that is executable by the processor. Upon execution by the processor, the regression analysis tree program operates to build a tree in accordance with the at least one user input by using a modified forward stepwise regression process to select the attributes for the tree'"'"'s branches from a plurality of attributes. After the tree is built, the regression analysis tree program then performs regression analysis to calculate at least one regression result for an attribute subset in a tree branch when the tree branch is in compliance with at least one criterion and the attribute subset has not been previously analyzed.

Claims

1. A method for enabling optimization of a parametric modeling process, the method comprising:

receiving a plurality of attributes;

allowing at least one user input;

building a tree in accordance with the at least one user input by using a modified forward stepwise regression process to select at least one attribute from the plurality of attributes; and

performing regression analysis to calculate at least one regression result for an attribute subset in a tree branch when the tree branch is in compliance with at least one criterion and the attribute subset has not been previously analyzed.

2. The method of claim 1, wherein allowing at least one user input comprises allowing a user to set at least one tree control.

3. The method of claim 2, wherein the at least one tree control comprises:

a limit on the number of levels of the tree; and

a limit on the number of nodes for each tree level.

4. The method of claim 3, wherein the limit on the number of nodes for each tree level varies by tree level such that each tree level is not limited to the same number of nodes.

5. The method of claim 1, wherein allowing at least one user input comprises allowing a user to define a data source, the data source including the plurality of attributes.

6. The method of claim 1, wherein allowing at least one user input comprises allowing a user to designate at least one suppressed attribute, the at least one suppressed attribute not being available for inclusion within the tree.

7. The method of claim 1, wherein allowing at least one user input comprises allowing a user to designate at least one suppressed part, the at least one suppressed part not being available for inclusion within the tree.

8. The method of claim 1, wherein allowing at least one user input comprises allowing a user to designate at least one selection criterion for selecting the regression results to be outputted to the user.

9. The method of claim 1, wherein allowing at least one user input comprises allowing a user to designate a sort order for outputting the regression results to the user.

10. The method of claim 1, wherein allowing at least one user input comprises allowing a user to extend regression results to another modeling tool.

11. The method of claim 1, wherein the tree building comprises filtering the plurality of attributes to determine available attributes for inclusion within the tree.

12. The method of claim 11, wherein filtering the plurality of attributes comprises:

removing duplicate attributes from the plurality of attributes;

removing poorly supported attributes from the plurality of attributes;

removing constant attributes from the plurality of attributes; and

removing attributes designated to be suppressed from the plurality of attributes.

13. The method of claim 11, wherein the tree building comprises:

calculating single variable regression formulas for each available attribute; and

using the single variable regression formulas to select at least one attribute for a first level of the tree.

14. The method of claim 13, wherein:

calculating single variable regression formulas for each available attribute comprises calculating an objective function for each available attribute; and

the attributes selected for inclusion within the first tree level have the highest objective function results.

15. The method of claim 14, wherein:

calculating an objective function for each available attribute comprises calculating an F-score for each available attribute; and

the attributes selected for inclusion within the first tree level have the highest F-scores.

16. The method of claim 13, wherein the tree building, for each tree level subsequent to the first tree level, comprises:

determining attribute candidates for a given node by;

removing from the available attributes each attribute previously selected for nodes on tree levels preceding the given node'"'"'s tree level; and

removing from the available attributes each attribute not in compliance with at least one criterion;

calculating regression formulas for each attribute candidate for the given node; and

using the regression formulas to select an attribute from the attribute candidates for the given node.

17. The method of claim 16, wherein:

calculating regression formulas for each attribute candidate for the given node comprises calculating an objective function for each attribute candidate; and

using the regression formulas to select an attribute from the attribute candidates for the given node comprises selecting the attribute candidate having the highest objective function result.

18. The method of claim 17, wherein:

calculating an objective function for each attribute comprises calculating an F-score for each attribute; and

selecting the attribute candidate having the highest objective function result comprises selecting the attribute candidate having the highest F-score.

19. The method of claim 16, wherein removing from the available attributes each attribute not in compliance with at least one criterion comprises removing from the available attributes each attribute having a correlation with an attribute previously selected in the given node'"'"'s branch higher than a threshold value.

20. The method of claim 1, wherein the at least one criterion comprises a minimum coefficient of multiple determination threshold value.

21. The method of claim 1, further comprising performing additional regression analysis on the attribute subset when the attribute subset is in compliance with at least one criterion.

22. The method of claim 1, wherein performing additional regression analysis on the attribute subset comprises calculating a maximum predicted to actual ratio.

23. The method of claim 1, wherein performing additional regression analysis on the attribute subset comprises calculating a stacking index.

24. The method of claim 1, further comprising logging the regression results.

25. The method of claim 1, further comprising using at least one regression formula of the tree to predict a price for a negotiation.

26. A system for enabling optimization of a parametric modeling process, the system comprising:

a processor;

an interface for allowing at least one user input; and

a regression analysis tree program executable by the processor, the regression analysis tree program operating to;

build a tree in accordance with the at least one user input by using a modified forward stepwise regression process to select at least one attribute from the plurality of attributes; and

perform regression analysis to calculate at least one regression result for an attribute subset in a tree branch when the tree branch is in compliance with at least one criterion and the attribute subset has not been previously analyzed.

27. The system of claim 26, further comprising a query and reporting facility for filtering, selecting, ranking, and sorting the regression results.

28. The system of claim 26, wherein the interface comprises an interactive screen.

29. Apparatus for enabling optimization of a parametric modeling process, the apparatus comprising:

a module for implementing an interface, the interface allowing at least one user input;

a module for receiving a plurality of attributes;

a module for building a tree in accordance with the at least one user input by using a modified forward stepwise regression process to select at least one attribute from the plurality of attributes; and

a module for performing regression analysis to calculate at least one regression result for an attribute subset in a tree branch when the tree branch is in compliance with at least one criterion and the attribute subset has not been previously analyzed.

30. The apparatus of claim 29, wherein the interface comprises at least one interactive screen.

31. The apparatus of claim 29, further comprising a module for implementing a query and reporting facility, the query and reporting facility filtering, selecting, ranking, and sorting the regression results.