This question already has an answer here:

I want to find anomalies in a time series. Is it possible to find anomalies using moving average?


marked as duplicate by Rob Hyndman, Scortchi, Momo, Nick Cox, Glen_b Nov 4 '13 at 13:45

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • $\begingroup$ no, but with some running median it should be possible $\endgroup$ – user603 Nov 4 '13 at 11:31

Anomalies can be easily detected while using "moving averages" . Please review the woRk or Tsay http://www.unc.edu/~jbhill/tsay.pdf . You might also search for "AUTOMATIC DETECTION OF INTERVENTION VARIABLES" using Google . Post your actual data and I will post the results.

  • $\begingroup$ Hi sir, thank you for your answer, I have a very long time series (about 10000 integere values). Im am not sure, if I can attach the file here. Which method do you find better to find anomalies? FFT, Wavlet transform or moving average- Is there any difference between anmolay and outlier in time series? $\endgroup$ – TangoStar Nov 4 '13 at 12:42
  • $\begingroup$ Post the data to dropbox.com. Check out Ruey Tsay's paper on outliers www.unc.edu/~jbhill/tsay.pdf‎ $\endgroup$ – Tom Reilly Nov 6 '13 at 19:50

Not the answer you're looking for? Browse other questions tagged or ask your own question.