Deriving predictive importance networks

• US 20060179019A1
• Filed: 11/19/2004
• Published: 08/10/2006
• Est. Priority Date: 11/19/2004
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: Intel Corporation
• Inventors: Bradski, Gary R.
• Granted Patent: US 7,644,049 B2
• Time in Patent Office: Days
• US Class Current: 706/20
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/20 Ensemble learning

Abstract

For a first feature of a dataset having a plurality of features, training a classifier to predict the first feature in terms of other features in the data set to obtain a trained classifier; scrambling the values of a second feature in the data set to obtain a scrambled data set, executing the trained classifier on the scrambled data set, determining predictive importance of the second feature in predicting the first feature based at least in part on the accuracy of the trained classifier in predicting the first feature when executed with the scrambled data set and creating a graph of the data set in which each of the first and the second features is a node of the graph and a label on an edge between the first node and the second node is based at least in part on the predictive importance of the first feature in terms of the second feature.

Claims

1. A method comprising:

for a first feature of a data set having a plurality of features, training a classifier to predict the first feature in terms of other features in the data set to obtain a trained classifier;

scrambling the values of a second feature in the data set to obtain a scrambled data set;

executing the trained classifier on the scrambled data set;

determining predictive importance of the second feature in predicting the first feature based at least in part on the accuracy of the trained classifier in predicting the first feature when executed with the scrambled data set; and

creating a graph of the data set in which each of the first and the second features is a node of the graph and a label on an edge between the first node and the second node is based at least in part on the predictive importance of the first feature in terms of the second feature.

2. The method of claim 1 wherein the dataset comprises a plurality of data points, each data point having a value for each feature, and the accuracy of the trained classifier comprises the fraction that the number of data points in the scrambled data set for which the trained classifier correctly predicts the value of the first feature (denoted hereinafter by Np) is of the total number of data points in the data set (denoted hereinafter by Nt), (Np/Nt).

3. The method of claim 2 wherein scrambling the values of the second feature in the data set to obtain a scrambled data set further comprises randomly remapping the values of the feature from the data points of the original data set to the data points of the scrambled data.

4. The method of claim 3 wherein at least one of the features is a label.

5. The method of claim 3 wherein at least one of the features is a predictor.

6. The method of claim 3 wherein the classifier is a tree-based classifier.

7. The method of claim 3 wherein the classifier is a forest based classifier

12. The machine readable medium of claim 11 wherein the dataset comprises a plurality of data points, each data point having a value for each feature, and the accuracy of the trained classifier comprises the fraction that the number of data points in the scrambled data set for which the trained classifier correctly predicts the value of the first feature (denoted hereinafter by Np) is of the total number of data points in the data set (denoted hereinafter by Nt), (Np/Nt).

13. The machine readable medium of claim 12 wherein scrambling the values of the second feature in the data set to obtain a scrambled data set further comprises randomly remapping the values of the feature from the data points of the original data set to the data points of the scrambled data.

14. The machine readable medium of claim 13 wherein at least one of the features is a label.

15. The machine readable medium of claim 13 wherein at least one of the features is a predictor.

16. The machine readable medium of claim 13 wherein the classifier is a tree-based classifier.

17. The machine readable medium of claim 13 wherein the classifier is a forest-based classifier.

8. A system comprising:

a processor communicatively coupled to a storage device;

the storage device having stored thereon data which when accessed by the processor causes the processor to perform a method, the method comprising;

for a first feature of a dataset having a plurality of features, training a classifier to predict the first feature in terms of other features in the data set to obtain a trained classifier;

scrambling the values of a second feature in the data set to obtain a scrambled data set;

executing the trained classifier on the scrambled data set;

determining predictive importance of the second feature in predicting the first feature based at least in part on the accuracy of the trained classifier in predicting the first feature when executed with the scrambled data set; and

creating a graph of the data set in which each of the first and the second features is a node of the graph and a label on an edge between the first node and the second node is based at least in part on the predictive importance of the first feature in terms of the second feature.

9. The system of claim 8 wherein the dataset comprises a plurality of data points, each data point having a value for each feature, and the accuracy of the trained classifier comprises the fraction that the number of data points in the scrambled data set for which the trained classifier correctly predicts the value of the first feature (denoted hereinafter by Np) is of the total number of data points in the data set (denoted hereinafter by Nt, (Np/Nt).

10. The system of claim 9 wherein scrambling the values of the second feature in the data set to obtain a scrambled data set further comprises randomly remapping the values of the feature from the data points of the original data set to the data points of the scrambled data.

11. A machine readable medium having stored thereon data that when accessed by a machine causes the machine to perform a method, the method comprising:

for a first feature of a dataset having a plurality of features, training a classifier to predict the first feature in terms of other features in the data set to obtain a trained classifier;

scrambling the values of a second feature in the data set to obtain a scrambled data set;

executing the trained classifier on the scrambled data set; and

determining predictive importance of the second feature in predicting the first feature based at least in part on the accuracy of the trained classifier in predicting the first feature when executed with the scrambled data set; and

creating a graph of the data set in which each of the first and the second features is a node of the graph and a label on an edge between the first node and the second node is based at least in part on the predictive importance of the first feature in terms of the second feature.