How to make box, violin and beeswarm scatter plots

In the scatter template you can add box, violin and beeswarm plots to help show the distribution of your data.

Box plots

Box plots, also known as box-and-whisker diagrams, are a common and useful statistical visualization. They let you see the distribution of your data quickly by highlighting the median and quartiles of each category. Optional “whiskers” extend out from the boxes and can be used to show which data points are outliers.

1

In the Box, violin, and beeswarm plots settings in the Preview tab, select yes to show a box plot.

2

Whisker type

Choose whether to show whisker lines at the highest and lowest values (select all dots) or exclude outliers, through the Box, violin and beeswarm settings.

3

Max whisker length

If you have chosen to exclude outliers, this setting lets you choose set a maximum distance of how much they should extend out of the box – meaning that you can also remove them completely. As a rule of thumb, each whisker should not extend further than 1.5 times the IQR (interquartile range, in other words, the difference between the third minus the first quartile). Everything beyond this value will be considered an outlier.

Violin plots

Similar to a box plot, violin plots do a good job of showing both the overall distribution of a dataset and the position of each individual point. The distribution is drawn around the dots, often in a sort of “violin” shape.

1

In the Box, violin and beeswarm plots settings in the Preview tab, select yes to show a violin plot.

2

Bandwidth

By default Flourish will set the bandwidth for you based on your data. Adjusting the bandwidth will change the shape of the violin plot, with a lower bandwidth generating lumpier plots which can help identify smaller clusters of values.

Default bandwidth:

Low bandwidth

3

Number of samples

This setting controls the number of points at which to calculate a violin's distribution. More points means a more accurate curve but can also harm performance.

Beeswarm plots

Beeswarm plots are great when you want to show both the overall distribution of a variable and the individual data points. Unlike basic scatter plots, which display the relationship between two quantitative variables, beeswarms show the distribution of a single numeric metric across one or more categories. Instead of proximate dots overlapping like they might in a dot plot, in a beeswarm, the dots “swarm” together around the axis.

NOTE: Dot sizing is ignored on beeswarm plots.