|
Activity Number:
|
546
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Graphics
|
| Abstract - #309125 |
|
Title:
|
Characterizing Multivariate Data with High-Resolution Human Faces
|
|
Author(s):
|
Dean Nelson*+ and Joe Szurek
|
|
Companies:
|
University of Pittsburgh at Greensburg and University of Pittsburgh at Greensburg
|
|
Address:
|
1052 Route 259, Ligonier, PA, 15658,
|
|
Keywords:
|
Multivariate data ; High-resolution human faces ; Chernoff's faces ; Principal components
|
|
Abstract:
|
Over 30 years ago, Chernoff used cartoon images of faces to depict multivariate data. The development of technologies to capture and manipulate high-resolution images makes it unnecessary to limit our characterization of human faces to cartoons. An analogue of Chernoff's method can be applied directly to high-resolution images of human faces. Because this is done using a high dimension 'face space', transformations based on principal components can be used to characterize multivariate data with human faces that facilitate interpretation of a variety of multivariate statistical analyses (e.g., discriminant analysis, cluster analysis, and multidimensional scaling). Additional uses of the technique will also be presented.
|
