Dear all,
I am using the biwavelet package to run time series analysis, but I am
trying to calculate a 'robust' version of the 95% CI using simulated
distribution which I generate under special (non-linear) conditions.
Stated differently, the standard 'white' and 'red' noise spectra
normally used to calculate CIs in wavelet analysis are not appropriate
for my data, but I have a separate simulation process that generates a
distribution that is an appropriate null model.
So, I have a vector of observed (M) and simulated data (Si). I can
easily perform the wavelet transform on M and extract the power matrix
(Mp); similarly, it's also easy enough to iterate through Si and
generate a distribution of power matrices for Mp. But is this the right
approach? My question is, what is the best approach for calculating and
plotting the wavelet CI using my simulated distribution, rather than the
red or white noise spectrum?
Any help will be greatly appreciated.
-Sean
--
Sean S. Downey
Assistant Professor
Department of Anthropology
University of Maryland
1111 Woods Hall
College Park, MD 20742
USA
Office: 301-405-1427
Email: sdown...@umd.edu <mailto:sdown...@umd.edu>
_______________________________________________
R-sig-ecology mailing list
R-sig-ecology@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
