CFE 2020: Start Registration
View Submission - CMStatistics
B0750
Title: On Dantzig and Lasso estimators of the drift in a high dimensional Ornstein-Uhlenbeck model Authors:  Dmytro Marushkevych - University of Luxembourg (Luxembourg) [presenting]
Mark Podolskij - University of Luxembourg (Luxembourg)
Gabriela Ciolek - University of Luxembourg (Luxembourg)
Abstract: New theoretical results are presented for the Dantzig and Lasso estimators of the drift in a high dimensional Ornstein-Uhlenbeck model under sparsity constraints. The focus is on oracle inequalities for both estimators and error bounds for several norms. We show that Dantzig and Lasso estimators have optimal rates, proving the restricted eigenvalue property solely under ergodicity assumption on the model. We also demonstrate the results of numerical analysis to uncover the finite sample performance of the Dantzig and Lasso estimators.