Skip to main content
Cross Validated

You are not logged in. Your edit will be placed in a queue until it is peer reviewed.

We welcome edits that make the post easier to understand and more valuable for readers. Because community members review edits, please try to make the post substantially better than how you found it, for example, by fixing grammar or adding additional resources and hyperlinks.

3
  • $\begingroup$ I really appreciate your response, thank you! Do you have any recommendations or know of any resources I could use to help me get started on these computations? $\endgroup$ Commented Feb 28 at 17:28
  • $\begingroup$ The location can be done with an arctangent and is quite easy (mu = atan2(Comp2, Comp1) ). The circular variance is slightly more complicated. Some practical considerations are in this paper: bpspsychub.onlinelibrary.wiley.com/doi/full/10.1111/bmsp.12108 There is also a something known about the transformation directly from the PN components to circular standard deviation or kappa, I think I read about it in Directional Statistics by Mardia, but I'm not completely sure. $\endgroup$ Commented Mar 3 at 12:28
  • $\begingroup$ I should note, by the way, that the spread only depends on the distance of the two components from the origin, so just taking sqrt(Comp1^2 + Comp2) is already a good measure of spread, with larger values indicating higher concentration. I think there is a monotonic transformation of that distance to circular standard deviation metrics, which is what I was referring to that might be in the Mardia book. Also, note that $\kappa$ is a parameter of the von Mises distribution, so it does not necessarily make sense to transform PN components to that. $\endgroup$ Commented Mar 3 at 16:11