MULTIDIMENSIONAL SCALING AS A TOOL TO ANALYZE THE STRUCTURE IN DATASETS.

Statistical methods are increasingly relevant for uncovering structure in different datasets. This paper aims to explain one of these methods that is known as Multidimensional Scaling (MDS) by using simple and intuitive examples. Multidimensional Scaling accepts quantitative, qualitative and mix data. There is a wide variety of methods to obtain appropriate data for MDS. The most direct way is to ask subjects to give pairwise ratings or to sort stimulus objects according to their similarity, relatedness and association. There are three types of this method: classical MDS, metric MDS and non-metric MDS. The purpose of each is to provide a visual representation of the pattern of proximities among a set of objects in a low number of dimensions (Borg, I. & Groenen, F. J. F., 2005). [ABSTRACT FROM AUTHOR]