SOME CONFIDENCE INTERVALS FOR LARGE CONCENTRATION PARAMETER IN VON MISES DISTRIBUTION

Von Mises distribution, also known as the natural analog of the normal distribution on real line, has two parameters namely the mean direction and concentration parameter. This paper considers the problem related to finding the best confidence intervals (CI) for a large concentration parameter. Various CI are derived and their performances are evaluated via simulation studies. We found that the CI based on asymptotic distribution of ˆ  and CI based on Bootstrap-t Method are good in terms of coverage probability even though the CI based on distribution of mean direction and resultant length gives shorter expected length.