3

I'm currently running a default/basic notebook on EMR (Release label: emr-6.1.0, Applications: Spark 3.0.0, Zeppelin 0.9.0, JupyterHub 1.1.0) and i'm having some issues getting the notebook to output a data profiling report in HTML.

I've installed pandas-profiling a variety of ways, using custom bootstrap actions & with the command sc.install_pypi_package("pandas-profiling")

I run into the following issue after trying to generate the report's HTML using IPython. It only produces the object vs the HTML.

enter image description here

I'm also aware that adding %%local can help produce it, like below.

enter image description here

But installing through bootstrap actions/on the notebook does not install from wherever %%local is located. As seen below.

enter image description here

So my first question is, can this profile report be produced without the %%local magic? I know there is also an %%html magic command, but that cannot print out a variable based on my testing, hence why I need IPython.

Second question is, how can pandas-profiling be added to this %%local environment? Should this even be the approach?

Thanks!!

Tbailey20
  • 85
  • 6

0 Answers0