14

I am looking for good references on using directional data (measure of direction in degrees) as an independent variable in regression; ideally, it would also be useful for hierarchical nonlinear models (the data are nested). I am also interested in directional data more generally.

I have found a text by Mardia, which I am going to get, but wondered if there were good articles.

I am more interested in practical articles about how to deal with this type of data than in theorems and proofs, or formal statements of distributions and such. Thanks

UPDATE I have got the Mardia text, which is quite comprehensive. After some more reading, I may be back with more questions.

Peter Flom
  • 94,055
  • 35
  • 143
  • 276
  • 1
    See also https://stats.stackexchange.com/questions/225729/what-are-periodic-version-of-splines – kjetil b halvorsen Aug 16 '19 at 02:12
  • 1
    Thanks! I'm curious - how did you find my question from 8 years ago and remember to put this there? I mean, I appreciate it, but I'm amazed! – Peter Flom Aug 16 '19 at 11:31
  • 1
    From time to time (often) I just go through, say som old questions from some user, and upvotes, edits, flags, comments as needed, or maybe even answering ... and I think this site can become more useful with crossreferencing, and such comments are stored as crossreferences. – kjetil b halvorsen Aug 21 '19 at 12:34

1 Answers1

8

I would suggest applying a transform which deals with periodicity. i.e. $\lim_{x \to 360} f(x) = f(0)$. An easy option is to take the sin and cos, and put them both as covariates in the model.

Simon Byrne
  • 3,336
  • 15
  • 29