Bibliographic Metadata

Title
Confidence sets based on the adaptive Lasso estimator / Nicolai David Amann
Additional Titles
Konfidenzmengen basierend auf dem adaptiven Lasso Schätzer
AuthorAmann, Nicolai David
CensorSchneider, Ulrike
PublishedWien, 2018
Description30 Blätter : Diagramme
Institutional NoteTechnische Universität Wien, Diplomarbeit, 2018
Annotation
Abweichender Titel nach Übersetzung der Verfasserin/des Verfassers
LanguageEnglish
Document typeThesis (Diplom)
Keywords (EN)model selection / inference / confidence set / / Lasso
URNurn:nbn:at:at-ubtuw:1-115033 Persistent Identifier (URN)
Restriction-Information
 The work is publicly available
Files
Confidence sets based on the adaptive Lasso estimator [19.97 mb]
Links
Reference
Classification
Abstract (English)

This thesis deals examines the adaptive LASSO estimator in the setting of moving parameter in the low-dimensional case, while the tuning parameters may vary over the components. The main part deals with the construction of asymptotic confidence sets based on the adaptive LASSO estimator in the case where at least one component of the tuning parameter is tuned to perform consistent model selection. The asymptotic distribution of the appropriately scaled and centered adaptive LASSO estimator is derived implicitly as the minimizer of a stochastic function, which is used to create confidence sets with asymptotically infimal coverage probability of 1. Besides confidence sets of the partially consistent tuned adaptive LASSO estimator, a condition on the tuning parameters is shown to be equivalent to consistency in parameter estimation. Conditions concerning the consistency in model selection are also derived. In particular, obtaining consistency in model selection for the adaptive LASSO estimator requires consistency in parameter estimation.

Stats
The PDF-Document has been downloaded 12 times.