18 January 2010 Visualizing multidimensional data through granularity-dependent spatialization
Author Affiliations +
Proceedings Volume 7530, Visualization and Data Analysis 2010; 75300M (2010); doi: 10.1117/12.838430
Event: IS&T/SPIE Electronic Imaging, 2010, San Jose, California, United States
Abstract
Spatialization is a special kind of visualization that projects multidimensional data into low-dimensional representational spaces by making use of spatial metaphors. Spatialization methods face a dual challenge: on the one hand, to apply dimension reduction techniques in order to overcome the limitations of the representational space, and on the other hand, to provide a metaphoric framework for the visualization of information at different levels of granularity. This paper investigates how granularity is modeled and visualized by the existing spatialization methods, and introduces a new approach based on kernel density estimation and landscape metaphor. According to our approach, clusters of multidimensional data are revealed by landscape "relief", and are hierarchically organized into different levels of granularity through landscape "smoothness." In addition, it is demonstrated, herein, how the exploration of information at different levels of granularity is supported by appropriate operations in the framework of an interactive spatialization environment prototype.
© (2010) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Sofia Kontaxaki, Eleni Tomai, Margarita Kokla, Marinos Kavouras, "Visualizing multidimensional data through granularity-dependent spatialization", Proc. SPIE 7530, Visualization and Data Analysis 2010, 75300M (18 January 2010); doi: 10.1117/12.838430; http://dx.doi.org/10.1117/12.838430
PROCEEDINGS
12 PAGES


SHARE
KEYWORDS
Visualization

Information visualization

Dimension reduction

Prototyping

Principal component analysis

Distance measurement

Visual analytics

RELATED CONTENT


Back to Top