Interactive Exploration of Ingredient Mixtures Using Multiple Coordinated Views

The complex nature of multivariate data sets calls for high interactive performance and intuitive metaphors. A specific type of multivariate data is where the variables sum up to a constant, here defined as multicomponent data. This application paper presents an interactive application for analysis of modeled multicomponent data. The aim is to find high performance variable combinations that fulfill some requested properties. The application is based on coordinated views that include parallel coordinates, a ternary diagram, a 2D scatter plot and a line plot. It supports numerous interaction techniques enabling fast analysis of complex patterns in multicomponent data sets. The application is developed in collaboration with researchers within the fields of statistics and chemistry. An informal usability evaluation indicates that the interactive nature of the application clearly facilitates the analysis process.