How to implement secondary y-axis?

hvPlot
1

Hey there!
I’m trying to implement a secondary y-axis and really having no luck!

Easiest way to explain is with an example:

Video game ownership prop for top games
Video game ownership prop for top games905×493 21.1 KB

On the left y-axis I have absolute counts of video game ownership, and on the right y-axis I have the cumulative proportion. How can I implement a right y-axis on a different scale?

I’ve tried something like:

# INITIALIZE plots
plot_1 = df.hvplot.bar(x='game', y='count', yaxis='left')
plot_2 = df.hvplot.line(x='game', y='cumul_prop', yaxis='right')

# DISPLAY plots
plot_1 * plot_2

But no success :frowning:

Thank you for your time and please let me know if I’m not clear!

1 Like
2

Welcome.

It is not directly supported (yet). See Proposed API for twin-axis support · Issue #5355 · holoviz/holoviews · GitHub

The workaround is to use a hook, and it depends on what backend you use. This hook works but can give problems when using some of the more advanced features in Holoviews, which is the package hvplot uses behind the scene:

import hvplot.pandas
import numpy as np
import pandas as pd
from bokeh.models import GlyphRenderer, LinearAxis, LinearScale, Range1d

x = np.random.randint(10, size=(10, 3))
x[:, 2] = np.cumsum(x[:, 2], axis=0)

df = pd.DataFrame(x, columns=["game", "count", "cumul_prop"])

# INITIALIZE plots
plot_1 = df.hvplot.bar(x="game", y="count", yaxis="left")
plot_2 = df.hvplot.line(x="game", y="cumul_prop", yaxis="right")


def overlay_hook(plot, element):
    # Adds right y-axis
    p = plot.handles["plot"]
    p.extra_y_scales = {"right": LinearScale()}
    p.extra_y_ranges = {"right": Range1d(0, 50)}
    p.add_layout(LinearAxis(y_range_name="right"), "right")

   # find the last line and set it to right
    lines = [p for p in p.renderers if isinstance(p, GlyphRenderer)]
    lines[-1].y_range_name = "right"


(plot_1.opts(ylim=(0, 10)) * plot_2.opts(color="k")).opts(hooks=[overlay_hook])

image

A small note: If you need help next time, try to post a minimal, reproducible example (MRE) with imports and fake data. This will make it much easier for people to help you.

3 Likes
3

Update March 2025 in case anyone else has had this question and stumbles across this thread like I did…

Support for multiple axes has been implemented (Available HoloViews >= 1.17, requires Bokeh >=3.2) using overlay.opts(multi_y=True), as described in this example, which I copied here:

overlay = hv.Curve([1, 2, 3], vdims=['A']) * hv.Curve([2, 3, 4], vdims=['A']) * hv.Curve([3, 2, 1], vdims=['B'])
overlay.opts(multi_y=True)
1 Like