1

I have some angles and I am wondering how can I test whether the distribution of my data is normal or not. In fact, I want something similar to KS test. Howver, I want to use it for circular data to find out whether my circular data has normal (Gaussian) distribution or this hypothesis is rejected. How should I do this?

Aep
  • 151
  • 4
  • 1
    If your observations range from $-\pi$ to $\pi$, they do not follow a classical Gaussian. – Dave May 28 '21 at 03:20
  • @Dave. Can you please explain a little more? My angles are around specific values. For instance, the circular mean is about pi/2. I want to know whether the distribution is normal or not. Thanks in advance. – Aep May 28 '21 at 04:17

1 Answers1

0

Instead of testing whether the data is Normally distributed, you should test whether the data has a relevant Normal-like circular distribution, such as the von Mises or Projected Normal distributions.

You can find more about tests for those in this answer.

Kees Mulder
  • 1,414
  • 1
  • 10
  • 10
  • Thanks for your response. As far as I understood, I should create a von misses distribution with the same values of mean and kappa as my data and then use the Kuiper test to see whether these two are identical or not. Is this correct? – Aep May 28 '21 at 16:12
  • Kind of, but if you want to use these tests I recommend just looking up documentation around how to use the specific tests in the program you want to run them in. – Kees Mulder May 28 '21 at 20:06