Gatherplots: Generalized Scatterplots for Nominal Data

TL;DR

Gatherplots are an innovative extension of scatterplots designed to eliminate overplotting for nominal data, enabling clearer comparison of group sizes and compositions through stacked visual entities and interactive features.

Contribution

This paper introduces gatherplots, a novel visualization technique that improves over traditional scatterplots by reducing overplotting and enhancing interpretability of nominal data groups.

Findings

Users judged subgroup proportions more accurately with gatherplots.

Gatherplots improved speed in understanding data distributions.

Expert review validated the design and utility of gatherplots.

Abstract

Overplotting of data points is a common problem when visualizing large datasets in a scatterplot, particularly when mapping nominal dimensions to one of the scatterplot axes. Transparency, aggregation, and jittering have previously been suggested to address this issue, but these solutions all have drawbacks for assessing the data distribution in the plot. We propose gatherplots, an extension of scatterplots that eliminates overplotting, particularly for nominal variables. In gatherplots, every data point that maps to the same position coalesces to form a stacked entity, thereby making it easier to compare the absolute and relative sizes of data groupings. The size and aspect ratio of data points can also be changed dynamically to make it easier to compare the composition of different groups. Furthermore, several embedded interaction techniques support slicing and dicing the gatherplot by pivoting on particular dimensions, ranges, and values in the dataset. Our evaluation shows that gatherplots enable users from the general public to judge the relative portion of subgroups more quickly and more correctly than when using conventional scatterplots with jittering. Furthermore, a review conducted by a group of visualization experts evaluated and commented on the gatherplot design.