Analysis and Assessment of Boxplot Characters for Extreme Data

The robust procedure used in constructing boxplot makes it to remain a vital tool for the display of distributional summaries, with no or less deviation from the empirical model characters which the data possess. In this paper, we investigate the embedded characters of the extreme dataset as richly displayed by a boxplot. We discuss and assess boxplot characters such as; the display of asymmetry, the outliers cut off using the outside rate per sample for three different types of boxplot implementations. The performance of the three boxplot fence implementations on extreme data was further assessed by introducing a new measure called fence sensitivity ratio. The fence sensitivity ratio is an attempt to propose an alternative to the conventional routine of data contamination in assessing the boxplot outlier rules. The findings in this paper highlighted the significance of boxplot as an exploratory data analysis tool in diagnosing some extreme data modelling tools and stress on the weakness of the existing boxplot methods and recommend useful suggestion in addressing such weaknesses for further investigation.