Overplotting, Jitter, and Sina Plots
Marco Cremonini · 2024
The easiest way of dealing with overplotting is by adjusting transparency of markers, this way the lighter or darker shade communicates a lesser or higher density. However, transparency alone might be sometimes enough to give a clear information, but insufficient in many cases. Another approach exists and it is called jittering. The idea is to introduce a tiny random error in the placement of markers in order to plot most of them visibly distinct. A sina plot is a relatively recent type of graphic, which attracted a good deal of interest because it is not just a simple variant of the basic jittering but it could be considered an alternative visualization to violin plots. A different approach for handling the overplotting problem, especially for categorical data, is provided by the beeswarm plot. This is another relatively graphic type defined, for ggplot, in a custom package to replicate the corresponding swarm plot natively implemented by Seaborn.