|
Activity Number:
|
289
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #305513 |
|
Title:
|
Robust Diagnostics for Multivariate Mixed Continuous and Categorical Data
|
|
Author(s):
|
Tsung-Chi Cheng*+ and Atanu Biswas
|
|
Companies:
|
National Chengchi University and Indian Statistical Institute
|
|
Address:
|
64 Section 2 ZhiNan Road, Taiepi, 11605, Taiwan
|
|
Keywords:
|
maximum trimmed likelihood estimator ; multiple outliers ; robust Mahalanobis distance ; robust diagnostics
|
|
Abstract:
|
In this article, we apply the maximum trimmed likelihood (MTL) approach (Hadi and Luceno 1997) to obtain the robust estimators of multivariate location and shape, especially for data mixed with continuous and categorical variables. The forward search algorithm (Atkinson 1994) is adapted to compute the proposed MTL estimates. A simulation study shows the proposed estimator outperforms the classical maximum likelihood estimator when outliers exist in data. Real datasets also are used to illustrate the method and results of the detection of the outliers.
|
