Detection of outliers in the unreplicated linear circular functional relationship model via functional form

摘要

In this paper we consider the problem of outliers for the functional relationship model of circular variables by transforming the circular data to continuous or real line data set via complex form. The COVRATIO statistic is extended from the linear regression models to the proposed model to detect any possible outliers. The cut-off points are obtained and the power of performance is examined by simulation studies. The model is illustrated with an application to the analysis of wind direction data recorded by two different techniques and the detection procedure of outliers is implied.