Visualization is an important tool for data analysis. As data sets become more complex (more variables), so too do the visualization techniques required. We developed two new visualization techniques, Attribute Blocks and Blend Blocks, which expand the limits on the number of simultaneously visualized, continuously defined variables. We also developed a third univariate technique, Gradient Textures, and combined these and some other visualization methods into a single visualization program.;This program was used to conduct an initial evaluation of the visualization methods to determine their potential for complex data analysis tasks. Attribute Blocks was the favored method for multivariate visualizations. Both Attribute Blocks and Blend Blocks showed promise in bivariate visualizations. Gradient Textures was seen as an interesting method, but was non-intuitive. Traditional visualization methods, such as isolines and single variable displays, still have a major role to play in data analysis. Finally, a common theme throughout the user evaluations was the importance of interactive displays.
展开▼
