* If you want to update the article please login/register
Appropriate color mapping for categorical data visualization can greatly aid in the discovery of underlying data patterns and effectively bring out visual aesthetics. We present an effective way to automatically produce a coloring that mimics the reference's while still allowing classes to be clearly identified by using a categorical data visualization and a reference image. We obtain a color palette with a high degree of separation between the colors by showcasing popular and distinguishable hues from the image's color space. These colors are assigned to specific classes by an integer quadratic algorithm to increase the given chart's point distinctness while keeping the color spatial relationships in the source image. With a diverse range of new coloring appearances for the input data, we display results on various coloring tasks. After using our software, user feedback confirms that the system's capability in automatically producing desirable colorings that match the user's expectation when choosing a reference.
Source link: https://doi.org/10.1007/s41095-021-0258-0
The desire for creating fair clustering algorithms has risen. The main aim is to ensure that the output of a cluster algorithm is not biased either toward or against specific subgroups of the population. The Multicluster algorithm for clustering categorical data developed by Santos and Heras can be modified in order to improve the cluster's fairness. Of course, fairness and effectiveness trade-off, so an increase in the fairness criterion often results in a loss of classification accuracy.
Source link: https://doi.org/10.1007/s10100-022-00824-2
* Please keep in mind that all text is summarized by machine, we do not bear any responsibility, and you should always check original source before taking any actions