HDTreeV : a Multidimensional Visualization Tool

As access to high-performance computing has increased over the years, the scientific community has in turn sought to analyze increasingly complex data. However, discovering and understanding complex relationships in multidimensional data can be a daunting task. Traditionally, exploratory data analysis has been used to discover patterns and garner substantive understanding of data by emphasizing the use of graphical representation, but methods for visualizing data beyond two or three dimensions are rarely used because of the inherent limitations of Cartesian representation where variables are mapped to spatial dimensions. In order to confront the task of visualizing multidimensional data, it is necessary to draw from principles of human visual perception and cognition. It is well-known that humans are incredibly adept at categorizing the natural world by reducing large degrees of freedom down to “oak trees” or “maple trees”, for instance. The overall shape of the trees is the result of high-order relationships and interactions among individual components, yet even in the face of such complexity, a young child can make these distinctions. Lindenmayer systems use simple procedural algorithms to render realistic flora that can be categorized just as easily as natural flora. Knowing this, the goal of the current project is to exploit the human ability to easily categorize nature by assigning variables in a multidimensional data set to features of binary trees. A tool, HDTreeV (High Dimensional Tree Visualization), has been created to provide the user a way to go through a process of categorizing and remapping, in order to ascertain multidimensional relationships in the data. Contact: Jeffrey R. Spies Dept. of Psychology University of Notre Dame Notre Dame, IN 46556 Tel: 1-574-631-7675 Fax: 1-574-631-8883 Email: jspies@nd.edu