An investigation of Multidimensional Scaling with an emphasis on the development of an R based Graphical User Interface for performing Multidimensional Scaling procedures

This dissertation is centered around the development of a graphical user interface, using the R statistical programing language, for performing Multidimensional Scaling. This program is called the MDS-GUI. Multidimensional Scaling (MDS) is one of the groups of multivariate analysis techniques that is used for dimension reduction. In general, these methods of MDS can be viewed as the problem of constructing a map when given a set of interpoint distances. The graphical configuration is produced, usually in two or three dimensions, in such a way that objects of the data are represented by points, where the Euclidean distances between them optimally represents the given set of observed distances. The MDS-GUI was developed using a combination of R and the scripting language tcltk. The primary objective of its design was to provide a comprehensive range of MDS methods and analytical tools that are accessed in a point and click manner. The target user group of the software is therefore widely spread as no coding and only a little expertise on MDS is required for its use. The capabilities of the MDS-GUI are demonstrated with the use of three data sets. The first is the well known and well used Morse-Code data; the second is a synthetic microarray based data set; and the third concerns the nutritional contents of a group of the cereals from the Kellog's company. The program will be the first complete MDS based GUI for the R-Environment, and will also be the package that provides access to the widest range of MDS methods in R.

Grantor dc:publisher.institution

Department of Statistical Sciences

Year dc:date.issued

2012

Author dc:creator

• Timm, Andrew

Advisor dc:contributor.advisor

• Lubbe, Sugnet

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/6880

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/6880