Title: Empirical band analysis of nonstationary time series
Authors: Scott Bruce - George Mason University (United States) [presenting]
Cheng Yong Tang - Temple University (United States)
Robert Krafty - University of Pittsburgh (United States)
Abstract: Power spectra of time series processes are defined over a continuous range of frequencies. However, time series data contain only a finite number of observations, so we must consider collapsed measures of power within local frequency bands that partition the frequency space. Frequency bands are used widely in the scientific literature and are often selected by manual observation of waveforms generated from a specific type of signal under particular settings. A standardized, unifying approach is provided to constructing customized frequency bands for different signals under study across different settings. A frequency-domain, iterative cumulative sum procedure is formulated to identify optimal frequency bands that best preserve nonstationary information. A formal hypothesis testing procedure is also developed to test which, if any, frequency bands remain stationary. This method is shown to consistently estimate the number of frequency bands and the location of the upper and lower bounds defining each frequency band.