Contrastive analysis for scatterplot-based representations of dimensionality reduction

Cluster interpretation after dimensionality reduction (DR) is a ubiquitous part of exploring multidimensional datasets. DR results are frequently represented by scatterplots, where spatial proximity encodes similarity among data samples. In the literature, techniques support understanding scatterplots’ organization visualizing importance features for cluster definition with layout enrichment strategies. However, current approaches usually focus on global information, hampering analysis whenever to understand differences clusters. Thus, this paper introduces methodology visually explore and interpret clusters’ formation based contrastive analysis. We also introduce bipartite graph relationship between statistical variables employed how influence formation. Our approach demonstrated through case studies, in which we two document collections related news articles tweets about COVID-19 symptoms. Finally, evaluate our quantitative demonstrate its robustness