Statistic for Outlier Detection in Circular Functional Relationship Model

Abstract

During the last few years, researchers have shown strong interest on the subject of outlier detection in both linear and circular for Error-in-Variables (EIV) Models. Recently, the studies of outlier detection on circular variables models using row deletion method are widely explored; in particular in regression and EIV models for circular variables. In this paper, we have proposed a new measure of mean circular error using cosine function for circular functional relationship model. We also used the row deletion method to detect observations that affect the measure the most, thus identifying them as outlier. The corresponding cut-off points are identified via simulation studies.