Some contributions to the spherical regression model
Parameters of some probability models in directional statistics are estimate by means of Bayesian and equivariant approaches. The authors provides the Bayes and equivariant estimators for the parameter of the von Mises-Fisher distribution on the Stiefel manifold. The methods developed are also used to estimate the orthogonal-parameter matrix in the spherical regression model. After defining a loss function on the circle and deriving a formula to compute the Bayes estimator, the Bayes estimator is obtained of the mean direction in the von Mises distribution for the cases where the concentration parameter is known and unknown. This risk analysis, empirical Bayes estimation of the mean direction, and Bayes prediction are also considered. Finally, two different methods are used to derived the approximate Bayes estimator of the concentration parameter.
- Research Organization:
- Kentucky Univ., Lexington, KY (United States)
- OSTI ID:
- 7173602
- Country of Publication:
- United States
- Language:
- English
Similar Records
Nonparametric conditional estimation