Try out Plot Composition in plotnine

[has2k1]

We're working on plot composition (inspired by patchwork) for the next release of plotnine (v0.15.0), and it's now available for early testing! This feature allows you to combine multiple plots into one by placing them side by side, or stacking them using a simple algebra.

Here is an example

from plotnine import *
from plotnine.data import mtcars

p1 = ggplot(mtcars) + geom_point(aes("wt", "mpg")) + labs(tag="a)")
p2 = ggplot(mtcars) + geom_boxplot(aes("wt", "disp", group = "gear")) + labs(tag="b)")
p3 = ggplot(mtcars) + geom_smooth(aes("disp", "qsec")) + labs(tag="c)")
p4 = ggplot(mtcars) + geom_bar(aes("carb"))

(p1 | p2 | p3) / p4

If you'd like to try it out before the official release, you can install the development version with:

pip install --pre plotnine

This will give you the latest v0.15.0.dev* v0.15.0a*~ version.

I'm especially looking for feedback as this feature evolves, feel free to share:

• Examples that worked (or didn't)
• Unexpected behavior or bugs
• Feature requests or improvements
• General impressions

Give it a spin and share your thoughts below 👇.

[PerryL-y]

Exciting features! Thanks for the work!

I just gave it a try. It seems that currently there's some bug when the ggplots have facet.

I have the following two ggplots with column facets:

p1

p2

combine them vertically works as expected

p1/p2

But horizontally combine them make it looks like:

p1+p2

[has2k1]

The size of the figure isn't big enough so the layout manager bails. If you increase the size of the last plot e.g. theme(figure_size=(10, 8)) or bigger, it should work.

There needs to be a warning message about what is happening.

[has2k1]

I just realised that you used the + operator above. Only | and \ are supported at the moment.

[openpaul]

Very nice, is there any documentation at all? Even if its just in code. I would love to test it, but I dont know exactly what to test from a first glance.
The code example you posted worked well for me btw. using 0.15.0-dev3

[openpaul]

I was playing around with the new composition. First of all: Very nice feature.

I was not able to figure out how to change the relative height of plots though if I have many rows.

Also I found one small bug where faceting will influence the label position poorly:

This works well:

from plotnine import *
from plotnine.data import mtcars

p1 = ggplot(mtcars) + geom_point(aes("wt", "mpg")) + labs(tag="a)")
p2 = ggplot(mtcars) + geom_boxplot(aes("wt", "disp", group = "gear")) + labs(tag="b)") + facet_grid("gear ~ .")
p3 = ggplot(mtcars) + geom_smooth(aes("disp", "qsec")) + labs(tag="c)")

(p1 | p2 | p3)

But here the labels are misaligned:

from plotnine import *
from plotnine.data import mtcars

p1 = ggplot(mtcars) + geom_point(aes("wt", "mpg")) + labs(tag="a)")
p2 = ggplot(mtcars) + geom_boxplot(aes("wt", "disp", group = "gear")) + labs(tag="b)") + facet_grid(". ~ gear")
p3 = ggplot(mtcars) + geom_smooth(aes("disp", "qsec")) + labs(tag="c)")

(p1 | p2 | p3)

Thanks again, and hope my report helps you make this composition feature even better.

[has2k1]

So far you cannot change the relative heights of rows or the relative widths of columns.

The misaligned tags have been fixed.

[ouzor]

Awesome, thanks! I had tried with patchworklib but it did make some silly mistakes. Tested with the pre-release version and it worked very nicely!