3

I was fascinated by the debugging and visualization library pixiedust developed by IBM and wanted to try it on gooogle colab.

I did following:

!pip install pixiedust

# matplotlib, seaborn, jinja2 are already installed in gcolab.

Display chart does not work

import pixiedust
df = pixiedust.sampleData() # this shows the dataframe
display(df)  # This does not do anything.

Debugging also does not work

import pixiedust

# Next cell
%%pixie_debugger
import random
def find_max (values):
    max = 0
    for val in values:
        if val > max:
            max = val
    return max
find_max(random.sample(range(100), 10))

# This gives nothing.

Is it possible to use pixiedust in Google Colab or it is only availabe in IMB clouds?

BhishanPoudel
  • 15,974
  • 21
  • 108
  • 169

1 Answers1

1

maybe its a bit late for @astro123, but perhaps someone else is also looking as I was. So here its my solution. As noted in the pixiedust install guidelines, at https://pixiedust.github.io/pixiedust/install.html you have to check you have a jupyter kernel with pixiedust. In colab just run

!jupyter pixiedust list

As you can see it does not have by default a kernel with pixiedust, just so install it by

 `   !jupyter pixiedust install`

Then you can select on the main panel to run with the pixiedust kernel. It should work now, but as collab is constantly evolving there is no a simple walkthrough.

Regards

  • Thanks Juan for your effort. Upvoted, but still, this does not work in colab. Do you have any public gcolab notebook that works for debugging? – BhishanPoudel Jun 30 '20 at 19:47
  • Well astro, I did not have one or the time now to check in deep, but as far as I have seen, colab is still around a 60% jupyter compatible, which means that, for instance automatic Slideshow without running Nbconverter, as the one offered by the nbextension called RISE did not worked on Colab. Check my additional answer for alternative sugestions, might one serve you correctly. Regards – Juan C. Vasquez B. Jul 01 '20 at 18:01