|
Activity Number:
|
277
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
IMS
|
| Abstract - #307006 |
|
Title:
|
Fractile Graphical Analysis with Multiple Covariates
|
|
Author(s):
|
Bodhisattva Sen*+
|
|
Companies:
|
University of Michigan
|
|
Address:
|
439 West Hall, Ann Arbor, MI, 48109,
|
|
Keywords:
|
fractile graphical analysis ; fractile graphs ; geometric quantile ; Mahalanobis quantile ; kernel regression
|
|
Abstract:
|
The concept of fractile graphical analysis (FGA) was introduced by Mahalanobis (1960). It is a method to compare the regression functions for two bivariate populations (X,Y). The method is particularly useful for comparing two regression functions where the covariates (X) for the two populations are not necessarily on comparable scales. Kernel-smoothed version of fractile regression functions (fractile graphs) are considered in Sen (Sankhya 2005). In this talk, we extend the notion of FGA to deal with the multiple covariate set up (when X is a vector). We develop smooth estimates of fractile graphs and study their statistical properties. In the process, we discuss suitable notions of multivariate quantiles (the geometric and Mahalanobis' quantile). The method has been applied to real data collected from the Reserve Bank of India to draw interesting and useful conclusions.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
