Elayouty, A., Scott, M. , Miller, C. and Waldron, S. (2018) Functional Principal Component Analysis for Non-stationary Dynamic Time Series. In: 33rd International Workshop on Statistical Modelling (IWSM 2018), Bristol, UK, 16-20 Jul 2018, pp. 84-89.
|
Text
169657.pdf - Accepted Version 515kB |
Abstract
Motivated by a highly dynamic hydrological high-frequency time series, we propose time-varying Functional Principal Component Analysis (FPCA) as a novel approach for the analysis of non-stationary Functional Time Series (FTS) in the frequency domain. Traditional FPCA does not take into account (i) the temporal dependence between the functional observations and (ii) the changes in the covariance/variability structure over time, which could result in inadequate dimension reduction. The novel time-varying FPCA proposed adapts to the changes in the auto-covariance structure and varies smoothly over frequency and time to allow investigation of whether and how the variability structure in an FTS changes over time. Based on the (smooth) time-varying dynamic FPCs, a bootstrap inference procedure is proposed to detect significant changes in the covariance structure over time. Although this time-varying dynamic FPCA can be applied to any dynamic FTS, it has been applied here to study the daily processes of partial pressure of CO2 in a small river catchment in Scotland.
Item Type: | Conference Proceedings |
---|---|
Status: | Published |
Refereed: | No |
Glasgow Author(s) Enlighten ID: | Waldron, Professor Susan and Scott, Professor Marian and Elayouty, Dr Amira and Miller, Professor Claire |
Authors: | Elayouty, A., Scott, M., Miller, C., and Waldron, S. |
College/School: | College of Science and Engineering > School of Geographical and Earth Sciences College of Science and Engineering > School of Mathematics and Statistics |
Copyright Holders: | Copyright © 2018 The Authors |
First Published: | First published in 33rd International Workshop on Statistical Modelling (IWSM 2018): 84-89 |
Publisher Policy: | Reproduced in accordance with the publisher copyright policy |
Related URLs: |
University Staff: Request a correction | Enlighten Editors: Update this record