|
Activity Number:
|
514
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #308840 |
|
Title:
|
Random Forests with Survey Data
|
|
Author(s):
|
Guillermo Mendez*+ and Sharon L. Lohr
|
|
Companies:
|
Arizona State University and Arizona State University
|
|
Address:
|
Dept of Mathematics and Statistics, Tempe, AZ, 85287-1804,
|
|
Keywords:
|
random forest ; importance scores ; clustered data
|
|
Abstract:
|
Random forests (Breiman, 2001) are a powerful tool for modeling data with both a categorical and continuous response. Although they are used primarily for prediction, a byproduct of the random forest algorithm is estimates of the importance scores of each predictor. One assumption for random forest, that the data in the training set is independently generated, will not necessarily hold when the data are clustered. We explore methods for estimating importance scores of potential variables for data from clustered samples.
|
