Colabtools: Better support for fast interactive plotting with HoloViews / hvPlot

[SandervandenOord]

There are currently 3 popular and easy ways to create fast interactive plots in Python with little code:

• Plotly Express -> Plotly 2 million downloads a month, although express only has 20k downloads
• Altair -> 130k downloads a month
• HoloViews / hvPlot (with the bokeh backend) -> 60k per month for Holoviews + 15k for hvPlot and both are on the rise.

Of course there's normal Plotly and Bokeh, but those are quite verbose and less suitable for fast plotting in my opinion.

I have two feature requests for easier plotting with HoloViews and hvPlot:
1) Currently, HoloViews and hvPlot are not pre-installed on Colab. It would be nice if this could be added. So I don't have to !pip install holoviews and !pip install hvplot every time.

2) Every time you want to show a HoloViews plot you have to add hv.extension('bokeh') to the cell in which you have your code for your plot. This is redundant. Hopefully this can also be fixed.

Example code:

# install holoviews and hvplot
!pip install holoviews
!pip install hvplot

# import + settings
import holoviews as hv
hv.extension('bokeh')

# plot simple line
# this unfortunately only works if I add hv.extension('bokeh') to every 
# cell in which I would like a plot. So it works, but it feels so redundant.
hv.Curve(range(0, 10))

Handy overview of all plotting tools and downloads: https://pyviz.org/tools.html

[jbednar]

See https://github.com/holoviz/holoviews/issues/3551 for background.

[korakot]

My hack to avoid hv.extension('bokeh') in every cell.

import holoviews as hv

def _render(self, **kw):
  hv.extension('bokeh')
  return hv.Store.render(self)
hv.core.Dimensioned._repr_mimebundle_ = _render

[tiagofassoni]

I'd like to point out that hvplot now needs bokeh>=2.0.0. Bokeh 2.0 needs tornado>5, which absolutely breaks with google-colab requirements. So using pip now doesn't work, at least without specifying version numbers.

[blois]

@tiagofassoni I'm seeing warnings on the install but the upgrade appears to work. Are you seeing something different?

https://colab.research.google.com/gist/blois/00748c23f85066f7b67c2672c241cec3/holoviews-install.ipynb

[korakot]

@bloise: Google colab uses tornado to start background server to serve local files, right?

It currently requires tornado v.4. Does it now also work with tornado 6 too? If so, please update requirement, so we can avoid error messages.

[zouhx11]

Holoviz is a better than plotly and altair, I don't know why colab doesn't integrate this wonderful visualization framework.

[tusharpoddar]

When I try to download holoviews I get an error. Is there a recent notebook that shows how to use holviews with google colab?

[korakot]

Later versions of holoviews has conflicting dependencies with Colab.

I test that 1.12.7 still works fine.

!pip install holoviews==1.12.7

[tusharpoddar]

When I install holoviews now there is no error bu then when I run this code I get an error - image at the end.

# Plotting interactive plots - 
import holoviews as hv
from holoviews import opts
hv.extension('bokeh', 'matplotlib')

import os
os.environ['HV_DOC_HTML'] = 'true'
#%env HV_DOC_HTML=true

import numpy as np
from bokeh.plotting import figure, show
from bokeh.io import output_notebook
hv.extension('bokeh')
output_notebook()
import imp
imp.reload(hv)

hv_spec = hv.Dataset(xr_spec)
hv.extension('bokeh')
hv_spec.to(hv.Image, ['time', 'freq'])

Error -

[korakot]

This code runs fine. You may have a bug somewhere else in your code.

!pip install -q holoviews==1.12.7
import holoviews as hv
%env HV_DOC_HTML=true
# select bokeh and plot
hv.extension('bokeh')
hv.Curve([1, 2, 3])

[tusharpoddar]

This code runs fine. You may have a bug somewhere else in your code.

!pip install -q holoviews==1.12.7
import holoviews as hv
%env HV_DOC_HTML=true
# select bokeh and plot
hv.extension('bokeh')
hv.Curve([1, 2, 3])

I am sorry to bother you, but I tried running this code but it get this error. Is there any other thing that you changed in your colab notebook?

For some reason, it gives me an error in the extension statement.

[korakot]

Restart the notebook and run again.

MENU > Runtime > Factory reset runtime

[tusharpoddar]

Restart the notebook and run again.

MENU > Runtime > Factory reset runtime

Thanks a lot for the help. It works now :)