Abstract
The least absolute deviation (LAD) estimator is an alternative to the OLS estimator, when outliers exist or the errors are distributed according to a heavy-tailed distribution. In this paper, we propose a shrinkage-type LAD estimator to improve the LAD one in the sense of accurate prediction. Our strategy is to apply an auxiliary information in estimation obtained from employing the LAD-LASSO, which gives the positive rule Stein-type shrinkage estimator as a result. A real data analysis confirms that our proposed estimator performs better in prediction error sense compared to the well-known LAD and LAD-LASSO regression estimator.
Type
Publication
1st Seminar on Nonparametric Statistics and its Application
