Holoviews change datashader colormap

Question

I'm trying to change the colormap used by datashader.

I tried this:

datashade(scatter, cmap='Reds')

Where scatter is an hv.Scatter element. This doesn't work because datashader expects an iterable or a function that returns colors. As such, this works:

datashade(scatter, cmap=['blue'])

So how can I take the 'Reds' colormap and convert it into something datashader can use?

Thank you.

Answer 1

Right; you can't pass the string name of a colormap to Datashader's cmap argument, because Datashader interprets a single string as the name of a single color, constructing a colormap from it by setting the R,G,B channels to that color and then varying the alpha channel. If you want a colormap, either pass a list of colors (as used by Bokeh for its palettes) or a Matplotlib colormap object (not the string name) to cmap:

from matplotlib import cm
datashade(scatter, cmap=cm.Reds)

Answer 2

To choose from any colormap available to Holoviews, use the following code:

from holoviews.plotting.util import process_cmap

datashade(scatter, cmap=process_cmap("Magma", provider="bokeh"))

Change "Magma" and "Bokeh" to any of the supported colormaps (Thanks to @Chris for the link).

---

Request: It would be nice if opts parameters given to Holoviews plots were automatically transferred if datashade supports them, as requested by this GitHub issue: https://github.com/holoviz/holoviews/issues/4125.

Answer 3

You can also use library hvplot which is built on top of HoloViews to create your plots, use datashader and change the color mapping, all in a convenient way:

import numpy as np
import pandas as pd
import hvplot.pandas

df = pd.DataFrame({
    'x': np.random.normal(size=100000),
    'y': np.random.normal(size=100000),
})

# use keyword datashade=True to turn on datashading
# use keyword cmap to change the default colormap
df.hvplot.scatter(
    x='x', 
    y='y', 
    datashade=True, 
    cmap='Magma',
)

Colormaps can be found here:
http://holoviews.org/user_guide/Colormaps.html

Resulting plot: