Use box plots/bubble plots for categories in scatterplot tab

[ShrimpCryptid]

Use Case

When examining categorical data in the scatterplot, the resulting graphs are difficult to parse because they give no density information per category.

Rendering with a box plot instead of a scatter plot would show more information, like median, density distribution, and potential outliers. See https://plotly.com/javascript/box-plots/ for more details.

(This could also be extended when comparing two categorical features to show markers/bubbles of varying sizes, as shown here: https://plotly.com/javascript/bubble-charts/).

Acceptance Criteria

• Replace the scatter plot with a box plot when one or more features are categorical.
• Optionally, if both features are categorical, show a bubble plot.

[ShrimpCryptid]

• Is it okay for the plot to automatically switch to box and whisker plots? Is there ever a situation where the scatterplot points are necessary?

[ShrimpCryptid]

Feedback from Chantelle: The scatter plot view is useful even if the points are merged together, because you can see where the currently selected track is in the population. If box plots were shown, they should be behind the main set of points and not block interaction with them or viewing.

Alternatively, Chantelle suggests adding spread to points along the categorical line or using a violin plot instead of the box plot.

[ShrimpCryptid]

Examples with jitter: https://community.plotly.com/t/how-to-show-overlap-points-in-scatter-plot/24148/6