Problem enter image description hereHave a time series of data and want to look at changes over long periods so want to use overlapping changes, then want to compare these overlapping changes to overlapping changes from a normal distribution in a QQ plot.

Questionso far I have had to resort to creating overlapping data via Monte Carlo from a normal distribution to use as the comparison. Is there a more deterministic/better way to do this?

Example of overlapping changes - green cel is change made up of difference from two yellow cells. It is overlapping as there is one calculated for each day.

  • 2
    $\begingroup$ Could you please describe for us what "overlapping changes" and "overlapping data" are? Perhaps you could give a small example. $\endgroup$ – whuber Jul 5 '17 at 22:18
  • $\begingroup$ Thanks for comment. Have edited and added an example. Hopefully that makes it clearer. $\endgroup$ – James65 Jul 6 '17 at 21:02
  • $\begingroup$ These are called "(lagged) differences." It would be unusual for your proposed analysis to yield much meaningful information--normality (or lack thereof) of these differences is typically way down on the list of things to examine. What are you actually trying to learn from your time series? $\endgroup$ – whuber Jul 6 '17 at 21:05
  • $\begingroup$ Am looking at different amounts of lag and observing visually as increase the lag how much closer it gets to normality. At first I tried using the discretisation/plotting points as (n-0.5)/N but realised with lagged overlapping differences that even a normal distribution was not straight line if lag is large. Therefore I had to resort to using simulated overlapping lagged differences. Just hoping there was a better way than that. $\endgroup$ – James65 Jul 6 '17 at 21:44

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.