[AlgebraOfGraphics.jl] adding jitter to a scatter plot points

Specific Domains Visualization
1

Hi,

I was wondering if anyone knows how (or if it is currently possible) to add jitter to the position of the points in this scatter plot generated with AlgebarOFGraphics.jl?

using CairoMakie, AlgebraOfGraphics, DataFrames, PalmerPenguins;

df = PalmerPenguins.load() |> DataFrame |> dropmissing;
	data(df) *
		mapping(:species, :bill_length_mm; color=:sex => nonnumeric) *
		visual(Scatter; markersize=10, alpha=0.5) |> draw

image
image800×600 25 KB

2
3

You could use RainClouds instead of Scatter. With plot_boxplots=false and clouds=nothing I think that is essentially what you want, but those options can be nice too. See rainclouds · Makie for the keyword arguments.

1 Like
4

Thanks for the suggestion. I wasn’t aware of the raincloud option in Makie. I’ll give it a try.

5

It’s pretty good! With

data(df) * mapping(:species, :bill_length_mm; color=:sex => nonnumeric) *
                       visual(RainClouds) |> draw

I get

rainclouds
rainclouds800×600 29.4 KB

and to get just jittered points,

data(df) * mapping(:species, :bill_length_mm; color=:sex => nonnumeric) *
                       visual(RainClouds; clouds=nothing, plot_boxplots=false) |> draw

plt
plt800×600 19.2 KB

Can also do clouds=hist:

data(df) * mapping(:species, :bill_length_mm; color=:sex => nonnumeric) *
                       visual(RainClouds; clouds=hist, plot_boxplots=false) |> draw

plt
plt800×600 21.6 KB

6

For what it’s worth, I think that these suggested plots look much better than just adding jitter to the original plot.

1 Like
7

Agreed! The histogram, boxplot, and density make this much more useful.

1 Like
8

Thanks again @ericphanson! I was actually able to generate the exact plot I was envisioning in my head with your RainCloud suggestion. Looks like the jitter functionality is tucked away in that function. Although I think the additional options the RainCloud function provides are much more useful, posting the jitter specific code here for others who might be looking for this type of plot in the future.

image
image800×600 41.8 KB

data(df) * mapping(:species, :bill_length_mm; color=:sex => nonnumeric) *
                       visual(RainClouds; markersize=10, jitter_width=0.25, alpha=0.5, clouds=nothing, plot_boxplots=false) |> draw
1 Like