0

I have time series of events. I observe that variance of inter-events intervals increases over the time.

Is it possible to detrend my time series?

Update:

Example of data is here. It is times when I observe event. Inter-events intervals are depicted below:

enter image description here

Just now I have the following solution:

1) I transform intervals($I$) to $log_{10}(I)$ 2) Then I simply detrend $log_{10}(I)$

Is it correct approach? Could you propose something more smart?

zlon
  • 639
  • 4
  • 20
  • 1
    It should be possible. Could you include a sample time series that would illustrate what kind of data you have? – Richard Hardy Feb 26 '17 at 16:51
  • Are you using the term "time series" [in its usual sense](http://stats.stackexchange.com/questions/126791)? It sounds like you might have a stochastic point process rather than a time series. – whuber Feb 26 '17 at 17:38
  • Some time series are a realisation of stochastic processes (of either kind) – zlon Feb 26 '17 at 18:50
  • It is from your link. I observe events during the time. LED blinking for example. It is stochastic processes. But as I understand I observe time series. – zlon Feb 26 '17 at 18:52
  • @RichardHardy I'he update post to answer Your question. – zlon Feb 27 '17 at 09:22

0 Answers0