Skip to content

wavelet analyisis to denoising time series

2 messages · intertodd, Wildi Marc (wlmr)

#
Hi everyone,

Im trying to use wavelet analysis to denoising time series data(stock index)
in order to perform a SVM forecast..
like use  db4, sqtwolog,4 levels...

There are a few packages avaliable with R:
wavelets: A package of funtions for computing wavelet filters, wavelet
transforms and multiresolution analyses. 

waveslim: Basic wavelet routines for one-, two- and three-dimensional signal
processing. 

wavethresh: Wavelets statistics and transforms. 

wmtsa: Insightful Wavelet Methods for Time Series Analysis

can anyone tell me which package should I use or maybe some codes please...

thanks guys

Best Regards

Louis



--
View this message in context: http://r.789695.n4.nabble.com/wavelet-analyisis-to-denoising-time-series-tp4655006.html
Sent from the Rmetrics mailing list archive at Nabble.com.
#
Before you get involved too deeply into this application please consider that wavelets are not well-suited for the task you intend to perform (SVM forecasting).
-Wavelets are well-suited for a posteriori data-analysis in the middle of a time series
-They fail to behave well towards the sample boundaries (the last data point).
But you're free to allocate your time as you want, of course.

Marc