Search code examples
pythonpython-3.xplotlyplotly-python

Can a Plotly visualization show separate Legends for Color, Symbol, Size, etc.?

Like ggplot2, can we have separate legends for Color, Symbol, etc. for a Plotly Express visualization?

from pydataset import data
import plotly.express as px
mtcars = data('mtcars')
mtcars.am = mtcars.am.astype('category')
mtcars.gear = mtcars.gear.astype('category')

plt = px.scatter(mtcars, x = 'mpg', y='qsec', color ='disp', symbol = 'am', size = 'wt')
plt.show()

Here, no separate legends are generated.

Also, If one of the legends has continuous value strip, then other legends become invisible.

enter image description here

Now, compare this with GGPLOT (I have used python package plotnine):

ggplot(mtcars, aes(x = 'mpg', y='qsec', color ='disp', shape = 'am', size = 'wt')) + geom_point()

enter image description here

So, simple and complete. How can we get an output like this using Plotly?

Thank you in advance.

Solution

  • I think your latest attempt looks pretty good. And personally I don't see the need for the size of legend elements to reflect sizes in the figure itself as long as the details otherwise are clear. Here's a little setup to adjust your legend and colorbar:

    fig.layout.legend.y = 1.05
fig.layout.legend.x = 1.035
fig.layout.coloraxis.colorbar.y = 0.35

    enter image description here

    Complete code:

    from pydataset import data
import plotly.express as px
mtcars = data('mtcars')
mtcars.am = mtcars.am.astype('category')
mtcars.gear = mtcars.gear.astype('category')

fig = px.scatter(mtcars, x = 'mpg', y='qsec', color ='disp', symbol = 'am', size = 'wt')

fig.layout.legend.y = 1.05
fig.layout.legend.x = 1.035
fig.layout.coloraxis.colorbar.y = 0.35
fig.show()