Simple circular regression model assuming wrapped cauchy error

Studying the association between random variables is an important subject in statistical analysis. in the case of circular variables,where their ranges are bounded,special type of regression models is needed. this paper proposes a new simple linear model for circular variables with a wrapped cauchy error due to its desirable characteristics of having a heavy tailed. the maximum likelihood estimates for model parameters are obtained via an iterative procedure and the standard error of the estimates as well as their confidence intervals are obtained by bootstrapping methods. the properties of estimates are investigated via a simulation study,where the unbiasness,consistency and robustness are verified. for illustration purposes,a real data set contains the average monthly wind directions in two main cities in the gaza strip-palestine have been analyzed and modeled by the proposed method. © 2015 pakistan journal of statistics.