|
Activity Number:
|
512
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #309040 |
|
Title:
|
A Test of Equality of Spectra of Short Memory Independent Gaussian Time Series
|
|
Author(s):
|
Ravi Siddani*+
|
|
Companies:
|
University of Maryland, Baltimore County
|
|
Address:
|
5003 Westland Boulevard, Baltimore, MD, 21227,
|
|
Keywords:
|
Spectral density ; Autocovariance function ; Periodograms ; Whittle Likelihood ; Contiguity
|
|
Abstract:
|
We develop a test of equality of the spectral densities of short memory independent Gaussian time series. This problem is equivalent to the equality of autocovariance functions. The autocovariance appears nonlinearly in the normal likelihood and is difficult to tackle analytically. We propose a test based on the asymptotic independence of the periodograms, and approach our null hypothesis of equality of spectral densities by an increasing sequence of hypotheses of equality of spectra at the Fourier frequencies. We examine the performance of our test statistic as the sample size increases. Appealing to the contiguity of the Whittle likelihood (an explicit function of the spectral density) and the normal likelihood we show that our test statistic maintains the size for large sample sizes. We apply our methodology to a rain gauge data set from Melbourne, Florida.
|
