Sphericity Test for Multitaper Spectral Estimation Parameter Selection

• Activity Number: 24
• Type: Topic Contributed
• Date/Time: Sunday, August 3, 2014 : 2:00 PM to 3:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Joshua Pohlkamp-Hartt*+
• Companies: Queen's University
• Keywords: Spectrum Estimation ; Time Series ; Multitaper ; Sphericity

Abstract

Spectrum estimation is a key tool for understanding the underlying processes governing time dependent data. One of the leading methods for performing this task is multitaper spectrum estimation. Within the multitaper method we must provide choices for the bandwidth parameter and number of tapers used. This requirement can lead to selection bias in the estimated spectrum and in turn can create misleading results for data analysis. In this paper we develop a test to determine if the parameter selection used is appropriate and give a methodology for determining an acceptable selection when a naïve choice is needed. We will show that the sphericity of the distribution of the residuals from the multitaper F-test is an indicator for the appropriateness of the parameter selection. We will evaluate this sphericity on simulated data and demonstrate that the method is effective at making a proper choice of parameters.