hyla07081963
11-06-2003, 03:18 PM
Hello,
I would like to use the "Lombs Algorithm" on short series of unevenly sampled data (heart rate data, one to three minutes) for spectral analysis.
Do I have to use "windowing" on the data ?
What if I don't ?
Concerning the Nyquist problem: The sample program already gives me a spectrum up to the "average Nyquist frequency". Do I still have to filter my interval series before doing the Lombs calculations ? How ?
And considering the "other end" : It is recommended not to use frequencies below 1/period ( e.g. with two minute periods 0,00833 hz) although the demo implementation does give these values. Is it viable (or necessary?) to cut the spectrum at these theoretical limits?
Greatly expecting Your answers,
Christoph
I would like to use the "Lombs Algorithm" on short series of unevenly sampled data (heart rate data, one to three minutes) for spectral analysis.
Do I have to use "windowing" on the data ?
What if I don't ?
Concerning the Nyquist problem: The sample program already gives me a spectrum up to the "average Nyquist frequency". Do I still have to filter my interval series before doing the Lombs calculations ? How ?
And considering the "other end" : It is recommended not to use frequencies below 1/period ( e.g. with two minute periods 0,00833 hz) although the demo implementation does give these values. Is it viable (or necessary?) to cut the spectrum at these theoretical limits?
Greatly expecting Your answers,
Christoph