Mean shift detection under long-range dependencies with ART

Atheoretical regression trees (ART) are applied to detect changes in the mean of a stationary long memory time series when location and number are unknown. It is shown that the BIC, which is almost always used as a pruning method, does not operate well in the long memory framework. A new method is developed to determine the number of mean shifts. A Monte Carlo Study and an application is given to show the performance of the method.

MoreLess

Year of publication:	2010
Authors:	Willert, Juliane
Publisher:	Hannover : Leibniz Universität Hannover, Wirtschaftswissenschaftliche Fakultät
Subject:	Zeitreihenanalyse \| Maßzahl \| Regression \| Theorie \| long memory \| mean shift \| regression tree \| ART \| BIC

freely available

Full text |

More access options

Check Google Scholar

In German libraries (KVK)

I need help

More details

Series:	Diskussionsbeitrag ; 437
Type of publication:	Book / Working Paper
Type of publication (narrower categories):	Working Paper
Language:	English
Other identifiers:	618572090 [GVK] hdl:10419/37121 [Handle] RePEc:han:dpaper:dp-437 [RePEc]
Classification:	C14 - Semiparametric and Nonparametric Methods ; C22 - Time-Series Models
Source:	EconStor

Persistent link: https://www.econbiz.de/10010270050