Predictor variable selection and dimensionality reduction for a predictive model

• US 8,751,273 B2
• Filed: 05/26/2010
• Issued: 06/10/2014
• Est. Priority Date: 04/16/2004
• Status: Active Grant

• Current Assignee: Longhorn Automotive Group LLC (Alpha Alpha Intellectual Partners LLC)
• Original Assignee: Brindle Data LLC (Intellectual Ventures LLC)
• Inventors: Pinto, Stephen K., Mansfield, Richard J. W., Rubin, Donald B., Hirshberg, Jay C.
• Primary Examiner(s): Jarrett, Scott L
• Application Number:
  US12/787,956
• Publication Number:
  US 20110071956A1
• Time in Patent Office: 1,476 Days
• Field of Search: 705/7.11
• US Class Current: 705/7.11
• CPC Class Codes: G06F 2111/08 Probabilistic or stochastic CAD
• CPC Class Codes: G06F 30/00 Computer-aided design [CAD]
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
• CPC Class Codes: G06Q 10/063 Operations research, analysis or management
• CPC Class Codes: G06Q 10/0633 Workflow analysis
• CPC Class Codes: G06Q 10/0635 Risk analysis of enterprise or organisation activities
• CPC Class Codes: G06Q 10/067 Enterprise or organisation modelling
• CPC Class Codes: G06Q 30/0202 Market predictions or forecasting for commercial activities

Abstract

Models are generated using a variety of tools and features of a model generation platform. For example, in connection with a project in which a user generates a predictive model based on historical data about a system being modeled, the user is provided through a graphical user interface a structured sequence of model generation activities to be followed, the sequence including dimension reduction, model generation, model process validation, and model re-generation.

In connection with a project in which a user generates a predictive model based on historical data about a system being modeled, and in which the project includes a series of user choice points and actions or parameter settings that govern the generation of the model based on rules, which direct the user to select and apply an optimal model.

Claims

1. A method for use by a computing system comprising one or more computing devices, the method comprising:

performing, by the computing system, based on an outcome variable selected for modeling, a cascade of a plurality of filtering operations, wherein the plurality of filtering operations comprises a determination of a first subset of a plurality of variables having a probability of non-contribution less than a first value and a determination, based on cross products of the first subset of the plurality of variables, of a second subset of the plurality of variables having a probability of non-contribution less than a second value;

determining, by the computing system, a subset of predictor variables from among the plurality of variables and a subset of excluded variables from among the plurality of variables;

generating, by the computing system, a sample predictive model comprising a model of the outcome variable based on variables in the subset of predictor variables;

receiving, by the computing system, a request to change the subset of predictor variables, wherein the request to change the subset of predictor variables comprises at least one of a request to add one of the subset of excluded variables to the subset of predictor variables, a request to add a new variable to the subset of predictor variables, and a request to remove a variable from the subset of predictor variables;

generating, by the computing system, a predictive model comprising a model of the outcome variable based on the variables in the changed subset of predictor variables, wherein the changed subset of predictor variables has fewer variables than the plurality of variables such that a dimensionality of the changed subset of predictor variables is different than a dimensionality of the plurality of variables; and

outputting, by the computing system, the generated predictive model.

2. The method of claim 1, further comprising:

displaying, by the computing system, a list of the subset of predictor variables and a list of the subset of excluded variables.

3. The method of claim 1, further comprising:

receiving, by the computing system, a request to change a dimensionality of a dataset.

4. The method of claim 3, wherein generating the predictive model comprises generating, by the computing system, the predictive model based on the request to change the dimensionality of the dataset.

5. The method of claim 1, further comprising:

receiving, by the computing system, a request to apply a variable transform to one or more of the variables in the subset of predictor variables.

6. The method of claim 5, wherein generating the predictive model comprises generating, by the computing system, the predictive model based on the request to apply the variable transform.

7. The method of claim 6, wherein the variable transform comprises one or more of a logarithm variable transform, a square variable transform, a square root variable transform, and an interaction variable transform.

8. The method of claim 1, the method further comprising:

partitioning, by the computing system, one of the subset of predictor variables into a plurality of predictive components;

wherein generating the predictive model comprises generating, by the computing system, the predictive model based on the plurality of predictive components.

9. A computer readable medium having instructions thereon for use by one or more computing devices, the instructions comprising:

instructions to perform, by the one or more computing devices, based on an outcome variable selected for modeling, a cascade of a plurality of filtering operations, wherein the plurality of filtering operations comprises a determination of a first subset of a plurality of variables having a probability of non-contribution less than a first value and a determination, based on cross products of the first subset of the plurality of variables, of a second subset of the plurality of variables having a probability of non-contribution less than a second value;

instructions to determine, by the one or more computing devices, a subset of predictor variables from among the plurality of variables and a subset of excluded variables from among the plurality of variables;

instructions to generate, by the one or more computing devices, a sample predictive model comprising a model of the outcome variable based on variables in the subset of predictor variables;

instructions to receive, by the one or more computing devices, a request to change the subset of predictor variables, wherein the request to change the subset of predictor variables comprises at least one of a request to add one of the subset of excluded variables to the subset of predictor variables, a request to add a new variable to the subset of predictor variables, and a request to remove a variable from the subset of predictor variables;

instructions to generate, by the one or more computing devices, a predictive model comprising a model of the outcome variable based on the variables in the changed subset of predictor variables, wherein the changed subset of predictor variables has fewer variables than the plurality of variables such that a dimensionality of the changed subset of predictor variables is different than a dimensionality of the plurality of variables; and

instructions to output, by the one or more computing devices, the generated predictive model.

10. The computer readable medium of claim 9, the instructions further comprising:

instructions to receive, by the one or more computing devices, a request to change a dimensionality of a dataset;

wherein the instructions to generate the predictive model comprise instructions to generate, by the one or more computing devices, the predictive model based on the request to change the dimensionality of the dataset.

11. The computer readable medium of claim 9, the instructions further comprising:

instructions to receive, by the one or more computing devices, a request to apply a variable transform to one or more of the variables in the subset of predictor variables;

wherein the instructions to generate the predictive model comprise the instructions to generate, by the one or more computing devices, the predictive model based on the request to apply the variable transform; and

wherein the variable transform comprises one or more of a logarithm variable transform, a square variable transform, a square root variable transform, and an interaction variable transform.

12. The computer readable medium of claim 9, the instructions further comprising:

instructions to partition, by the one or more computing devices, one of the subset of predictor variables into a plurality of predictive components;

wherein the instructions to generate the predictive model comprise instructions to generate, by the one or more computing devices, the predictive model based on the plurality of predictive components.

13. A system comprising:

at least one processor; and

a computer readable medium having instructions thereon, the instructions comprising;

instructions to perform, by the system, based on an outcome variable selected for modeling, a cascade of a plurality of filtering operations, wherein the plurality of filtering operations comprises a determination of a first subset of a plurality of variables having a probability of non-contribution less than a first value and a determination, based on cross products of the first subset of the plurality of variables, of a second subset of the plurality of variables having a probability of non-contribution less than a second value;

instructions to determine, by the system, a subset of predictor variables from among the plurality of variables and a subset of excluded variables from among the plurality of variables;

instructions to generate, by the system, a sample predictive model comprising a model of the outcome variable based on variables in the subset of predictor variables;

instructions to receive, by the system, a request to change the subset of predictor variables, wherein the request to change the subset of predictor variables comprises at least one of a request to add one of the subset of excluded variables to the subset of predictor variables, a request to add a new variable to the subset of predictor variables, and a request to remove a variable from the subset of predictor variables;

instructions to generate, by the system, a predictive model comprising a model of the outcome variable based on the variables in the changed subset of predictor variables, wherein the changed subset of predictor variables has fewer variables than the plurality of variables such that a dimensionality of the changed subset of predictor variables is different than a dimensionality of the plurality of variables; and

instructions to output, by the system, the generated predictive model.

14. The system of claim 13, the instructions further comprising:

instructions to receive, by the system, a request to change a dimensionality of a dataset;

wherein the instructions to generate the predictive model comprise instructions to generate, by the system, the predictive model based on the request to change the dimensionality of the dataset.

15. The system of claim 13, the instructions further comprising:

instructions to receive, by the system, a request to apply a variable transform to one or more of the variables in the subset of predictor variables;

wherein the instructions to generate the predictive model comprise the instructions to generate, by the system, the predictive model based on the request to apply the variable transform; and

wherein the variable transform comprises one or more of a logarithm variable transform, a square variable transform, a square root variable transform, and an interaction variable transform.

16. The system of claim 13, the instructions further comprising:

instructions to partition, by the system, one of the subset of predictor variables into a plurality of predictive components;

wherein the instructions to generate the predictive model comprise instructions to generate, by the system, the predictive model based on the plurality of predictive components.

17. A system comprising:

means for performing, by a computing system, based on an outcome variable selected for modeling, a cascade of a plurality of filtering operations, wherein the plurality of filtering operations comprises a determination of a first subset of a plurality of variables having a probability of non-contribution less than a first value and a determination, based on cross products of the first subset of the plurality of variables, of a second subset of the plurality of variables having a probability of non-contribution less than a second value;

means for determining, by the computing system, a subset of predictor variables from among the plurality of variables and a subset of excluded variables from among the plurality of variables;

means for generating, by the computing system, a sample predictive model comprising a model of the outcome variable based on variables in the subset of predictor variables;

means for receiving, by the computing system, a request to change the subset of predictor variables, wherein the request to change the subset of predictor variables comprises at least one of a request to add one of the subset of excluded variables to the subset of predictor variables, a request to add a new variable to the subset of predictor variables, and a request to remove a variable from the subset of predictor variables;

means for generating, by the computing system, a predictive model comprising a model of the outcome variable based on the variables in the changed subset of predictor variables, wherein the changed subset of predictor variables has fewer variables than the plurality of variables such that a dimensionality of the changed subset of predictor variables is different than a dimensionality of the plurality of variables; and

means for outputting, by the computing system, the generated predictive model.

18. The system of claim 17, further comprising:

means for receiving, by the system, a request to change a dimensionality of a dataset;

wherein the means for generating the predictive model comprises means for generating, by the system, the predictive model based on the request to change the dimensionality of the dataset.

19. The system of claim 17, further comprising:

means for receiving, by the system, a request to apply a variable transform to one or more of the variables in the subset of predictor variables;

wherein the means for generating the predictive model comprises means for generating, by the system, the predictive model based on the request to apply the variable transform; and

wherein the variable transform comprises one or more of a logarithm variable transform, a square variable transform, a square root variable transform, and an interaction variable transform.

20. The system of claim 17, further comprising:

means for partitioning, by the system, one of the subset of predictor variables into a plurality of predictive components;

wherein the means for generating the predictive model comprises means for generating, by the system, the predictive model based on the plurality of predictive components.