Estimation of regression parameters in the presence of outliers in the response |
| |
Authors: | Sugata Sen Roy Sibnarayan Guria |
| |
Institution: | 1. Department of Statistics , University of Calcutta , Calcutta, India ssrstat@caluniv.ac.in;3. Department of Statistics , Bidhannagar College , Calcutta, India |
| |
Abstract: | In this paper, we consider a regression model with non-spherical covariance structure and outliers in the response. The generalized least squares estimator obtained from the full data set is generally not used in the presence of outliers and an estimator based on only the non-outlying observations is preferred. Here we propose as an estimator a convex combination of the full set and the deleted set estimators and compare its performance with the other two. |
| |
Keywords: | outliers non-spherical covariance structure deletion technique convex combination |
|
|