I have a time series that is spotty (more 'no sample' than 'sample' at each second, which is the sample rate). There are obvious trends when viewed visually. However since there are holes, the FFT has a spike near zero. Is there anything that I can do about this?

  • $\begingroup$ Could you post a photo of your FFT plot? $\endgroup$ – Izzo Apr 15 at 22:08
  • $\begingroup$ what are you looking for in your FFT result? $\endgroup$ – Stanley Pawlukiewicz Apr 15 at 23:04
  • $\begingroup$ @StanleyPawlukiewicz I can see (visually) that the data is periodic by day and by season. Every day it peaks at noon, and early in the AM there is no data (no activity). So I'd like a spike at the day frequency, and at the month frequency. But my average is very small because of all the zeros (no data for that sample) $\endgroup$ – horse hair Apr 15 at 23:14
  • 2
    $\begingroup$ you could try Lomb-Scargle en.m.wikipedia.org/wiki/Least-squares_spectral_analysis $\endgroup$ – Stanley Pawlukiewicz Apr 16 at 0:33
  • $\begingroup$ Given you said that you have a spike near zero, do you mean near zero frequency? If at zero frequency then this is a result of DC Offset (non-zero mean), which can be eliminated simply by subtracting the mean. $\endgroup$ – Dan Boschen Apr 16 at 11:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.