Robustified total least squares

Result code in IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21340%2F10%3A00176584" target="_blank" >RIV/68407700:21340/10:00176584 - isvavai.cz</a>

Result on the web

—

DOI - Digital Object Identifier

—

Result language

angličtina

Original language name

Robustified total least squares

Original language description

Classical regression estimators, such as the ordinary least squares (LS), are sensitive to occurrence of outliers and are not consistent when the orthogonality condition fails. There have been several robust estimators that can cope with this problem. The development of instrumental weighted variables (IWV), the robust version of instrumental variables methods, is reviewed. The alternative approach in regression methods when orthogonality condition is breaking and both independent and dependent variables are considered to be measured with errors is called total least squares. The existence and uniqueness of the solution is discussed and different approaches of calculation are described. The robustified version of TLS based on the idea of downweightingthe influential points is presented and its properties are discussed. Finally the generalization of TLS to mixed LS-TLS and its robustified version is mentioned.

Czech name

—

Czech description

—

Type

D - Article in proceedings

CEP classification

BB - Applied statistics, operational research

OECD FORD branch

—

Project

—

Continuities

Z - Vyzkumny zamer (s odkazem do CEZ)

Publication year

2010

Confidentiality

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Article name in the collection

Doktorandské dny 2010

ISBN

978-80-01-04644-9

ISSN

—

e-ISSN

—

Number of pages

11

Pages from-to

—

Publisher name

Česká technika - nakladatelství ČVUT

Place of publication

Praha

Event location

Praha

Event date

Nov 19, 2010

Type of event by nationality

CST - Celostátní akce

UT code for WoS article

—