Robust Regression through Robust Covariances.
Abstract
This paper describes a regression estimator which builds on the investigation of the covariance structure in the full space of explanatory variables and response variable. It is robust since it down weights outlying observations. A comparison to other robust estimators is included. Keywords include: Bounded Influence; asymptotics, and simulation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1985
- Accession Number
- ADA153301
Entities
People
- R. Maronna
- S. Morgenthaler
Organizations
- Yale University