Activity Number:
|
551
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 3, 2016 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Royal Statistical Society
|
Abstract #318592
|
View Presentation
|
Title:
|
Dating Structural Breaks in Functional Data Without Dimension Reduction
|
Author(s):
|
Ozan Sonmez* and Alex Aue and Gregory Rice
|
Companies:
|
University of California at Davis and University of California at Davis and University of Waterloo
|
Keywords:
|
Change point analysis ;
Functional data ;
Functional time series ;
Functional principal components ;
Intra Day financial data ;
Structural breaks
|
Abstract:
|
An estimator for the time of a break in the mean of stationary functional data is proposed that is fully functional in the sense that it does not rely on dimension reduction techniques such as functional principal component analysis (fPCA). A thorough asymptotic theory is developed for the estimator of the break date for fixed break size and shrinking break size. The main results highlight that the fully functional procedure performs best under conditions when analogous fPCA based estimators are at their worst, namely when the feature of interest is orthogonal to the leading principal components of the data. The theoretical findings are confirmed by means of a Monte Carlo simulation study in finite samples. An application to one-minute intra-day cumulative log-returns of Microsoft stock data highlights the practical relevance of the proposed fully functional procedure
|
Authors who are presenting talks have a * after their name.