I have consecutive measurements of two different phenomena (x and y) on the same subject, giving me a collection of time-series x_ik(t) and y_ik(t), where i is the i-th measurement and k is the k-th subject. I would like to compare the variability across k of x and y to inquiry what of the two phenomena is more variable. Is there a formal way of doing it? Thanks!

EDIT: I figured that if I take the variance across i-s at each time sample, I run into a problem. The signal (x or y) that is more "steep" will have a higher variability because the variance is computed along the vertical axis. Is there a known way to solve this?

  • $\begingroup$ coefficients of variation or variance do not cut it for you? $\endgroup$ – Aksakal Jun 13 '18 at 15:41
  • $\begingroup$ I would certainly use the variance. I am just asking if there is a formal way of comparing variances in different time series. $\endgroup$ – Cristiano Jun 13 '18 at 15:53
  • $\begingroup$ Maybe run a time series decomposition on each of the time series and compare their remainders. Single-difference the time series data to put them on the same scale. $\endgroup$ – Brash Equilibrium Jun 13 '18 at 16:36

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.