Application of Contrast Driven Color-Class Assignment to Four Categorical Data Visualization Diagrams

In data visualization, when dealing with unordered categorical data, we create a variety of diagrams using a color-coded paradigm that links colors to data classes. If the chosen color palette contains a number of colors equal to the number of categories, the space search is reduced to permutations. This study aims to propose an algorithm that selects the ideal color assignment from a user-provided color palette for each category in unordered categorical data visualization. We enhance the viewer’s capacity to recognize geometric objects from different classes by applying a concept of contrast importance factors that expresses the necessity to get for a pair of object classes a strong color contrast to optimize legibility. The technique is based on a fitness function that separates on a first part, the need for color contrast using the diagram geometric data and, on the second part, the palette color distances. We show four categorical diagrams applications: streamgraphs, chord diagrams, polygonal maps and line-graphs. The article is an extension, showing applications to polygonal maps and line-charts, of a paper previously presented in Lisboa VISIGRAPP 2023 conference.