Advanced Interactions with Heatmaps for Analyzing High-Dimensional Datasets

We introduce an extended heatmap visualization integrated with enhanced interactive capabilities and computational tools for analyzing large and complex datasets. A heatmap is capable of displaying a large amount of data with many dimensions while minimizing graphic area. Varying levels of detail are provided by rendering cells as data values or computed values. Earlier approaches of having peer interfaces with R or other mathematical and analytic environments were powerful, but these were slower and lacked more interactive visual control and feedback. By including built-in computational tools, a variety of useful functions are made available to the user to enable quick identification of interesting patterns and relationships in a more guided exploratory environment. In a case study, we demonstrate the usefulness of this new visualization to solve problems related to drug discovery. Index Terms — Visual analytics, interaction techniques, large data visualization, multidimensional, multi-variate.