Time-varying Dynamic Functional Principal Components

Speaker: Amira El-Ayouty  (Glasgow University)

Abstract

The enormous technical advances in environmental sensors allow measurements to be recorded at a high temporal resolution over a long time period. Although these Environmental High-Frequency Data (EHFD) promote a deeper and broader understanding of the environmental systems, they present various challenges in terms of statistical modelling and analysis. The challenges arising in this context include: the large volumes of data, the strong correlation structure that persists over a large number of lags and the high variability and non-stationarity over the different timescales. Functional time series analysis and its recent developments in the frequency domain provide an appropriate tool for analysing such EHFD, taking into consideration the aforementioned technical challenges.