Outlier Labeling Methods for Medical Data

Outlier detection should be considered as preliminary step to avoid misinterpretation of results in data analysis. In this paper, the performance indices of the outlier labeling methods such as SD method, Median method, MADe method, Z−Score, Modified Z−Score, Tukey’s method for univariate data set were compared. Each labeling method has different measures and based on which an interval was constructed to detect outliers. We have attempted to find the appropriate choice of outlier detection methods. The advantages and disadvantages of each method were discussed.