Visual Analytics for Multidimensional Time-Varying Data via Dimension Reduced Visual Perception

In this paper, we propose a visualization system for multidimensional time-varying data analysis via dimension reduced visual perception, which enables users to better perceive the temporal features of the time-varying multidimensional data. Firstly, the orthogonal transformations are conducted to minimize the offsets of MDS plots between adjacent time steps, which do great favors for the visually tracking of temporal patterns of interest and maintaining an accuracy mental map. Then, hexagons are applied to tessellate the plane to eliminate the overlap in the resulting points and facilitate the visual perception and interaction of MDS plots. Furthermore, we employ a hierarchical clustering algorithm and design a specialized glyph to enhance the visual perception of temporal clusters. Finally, a cluster based grouping animation is designed to highlight the temporal patterns between adjacent timestamps. By integrating the above visualization methods, a time-varying multidimensional data visualization analysis prototype system is developed. We demonstrate the effectiveness and usefulness of TMDS in case studies with the economic data and air quality data.