This notebook breaks down how the author generated the dot density maps in his WashingtonPostarticle. Each dot represents 10 people mapped onto census block groups (the smallest geography used in the census).
This piece elaborates on best practices for presenting your insights virtually. We particularly like the suggestion around using a spotlighting pointer as well as the concept of building complex charts slowly.
When generating violin plots, this article suggests ordering by median value for clarity as well as adopting a horizontal layout when labels are long. It also explores when you should consider other graph types (e.g. ridgeline, boxplot).
This is a creative visualization of the output of NLP, with a log scale depicting how often either men or women's faces are described with a certain word.