[Zurück]

B. Pötscher, U. Schneider:
"Distributional Properties of Lasso-Type and Thresholding Estimators";
Vortrag: Econometrics Research Seminar, IHS, Wien (eingeladen); 15.12.2011.

Penalized least-squares estimators, such as the famous Lasso estimator, have been studied intensively in recent
literature. While many properties of these estimators are now well understood, the understanding of their distributional
characteristics, such as finite- and large-sample distributions, convergence rates and confidence sets is still incomplete.
We give a literature overview and discuss results for penalized least-squares estimators in an orthogonal linear
regression model. We also present results for thresholding estimators within a (non-orthogonal) linear regression model
where the number of parameters k can depend on sample size n and may diverge with n. In this settinge we study
versions of the estimators when the error-variance is unknown and also investigate the effects of having to estimate the
variance when the degrees of freedom n-k does not tend to infinity or does so very slowly.

http://publik.tuwien.ac.at/files/PubDat_204684.pdf