Univariate and bivariate distributions for axial and/or directional data
by
Barry C. Arnold
University of California , Riverside, USA
Coauthors: Ashis SenGupta

Observations on random directions, measured in radians, and thus constrained to take on values in the interval [0, 2p), are called directional data. Axial data, encountered when the observed random direction vectors do not have an identified starting point, are recognizable as directional data modulo p. We will survey univariate models (many of which are well known) for such data and will suggest some related (less well known) bivariate models, together with natural multivariate extensions.

Date received: July 21, 2006