An Improved Pseudo Distance Scale for Measuring One-to-many Relationships in Multidimensional Scaling

Abstract

In this study, we compared different pseudo distance measures used on a set of rainfall data recorded between the years 1968 and 2003 at nine rain gauge stations. The data is mapped on 2D plots where Euclidean distance and two different units of pseudoscale distance are applied through Multidimensional Scaling visualization techniques. The results show that a higher unit of Pseudo distance scale gives the smallest STRESS value when compared to the others. This implies that this type of distance measure reduces the mismatch between the distance rank order in the data and the rank order of distances in the ordinations. Keywords: ANOVA-Tukey; Euclidean-distance; Pseudo-distance; Multidimensional Scaling. 2010 Mathematics Subject Classification: 62H30; 65F35