1
$\begingroup$

I have a data series with approximatelly 300 values. I want to show that it is not weakly stationary, that is, show as mathematically as possible that its mean or its standard deviation vary with time. What would you recommend? I would like to be capable of implementing it on excel.

$\endgroup$
1
  • $\begingroup$ What did you do finally? Would be glad to know. $\endgroup$
    – keepAlive
    Oct 5, 2018 at 5:54

1 Answer 1

0
$\begingroup$

What about simply computing moving averages (MA) and moving variances (MV) so as to compare them side-by-side, respectively performing Welch's test for equality of means and F-test of equality of variances.

E.g. the variance over the period $t:t+n$ with the one over $t+\ell:t+n+\ell$, where $n\in(0,299-t-\ell]$ is the MV window size, $\ell\in(0,299-t-n]$ is the time distance between the two compared periods (possibly overlapping, but imperfectly), and $t\in[0,299−n-\ell]$. An illustration below

enter image description here

The F-test (1) as well as Welch's test (2) are easily doable with excel (see, e.g. (1) and (2)).

$\endgroup$
1
  • $\begingroup$ Any question @John ? $\endgroup$
    – keepAlive
    Oct 17, 2019 at 11:45

Your Answer

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

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