Conda Enviroments use in %python interpreter

Owned by Onesait Platform
Last updated: 19/03/24
3 min read

For the use of conda enviroments in the notebook module of notebooks, the %python.conda sub-interpreter is used with the following options:

  • Get conda information:

    %python.conda info

  • List all environments:

    %python.conda env list

  • Create a conda environment:

    %python.conda create --name [ENV NAME]

  • Activate an environment (the python interpreter is restarted):

  • Disable environment

  • Get the list of packages for the current environment

  • Install a package from conda repositories

  • Uninstall a package

For the use of conda environments in the notebook module of notebooks, the %python.conda sub-interpreter is used with the following options:

Creating the environment in a notebook

The first step is creating the conda environment.

%python.conda create --name py37 python=3.7

In the previous example we are creating a new environment based on Python 3.7 and named py37

Then we go on to activate it in order to install the necessary libraries:

%python.conda activate py37

For example, we try to import the pandas library

%python
import pandas

We verify that since it is not installed it will give us an error, so we launch its installation from conda

%python.conda install pandas

And when you finish the previous step, it is now possible to import the library correctly

Finally, we deactivate the environment

%python.conda deactivate

Python environment versions on notebooks

Keep in mind that there are limitations to the version of python that can be used (and therefore created in an environment).

For version zeppelin 0.8.2 (onesait platform < 5.0.0) → the maximum python version is 3.7

For zeppelin versions 0.10.1 (onesait platform >= 5.0.0) → the maximum python version is 3.9

Using an environment type in a notebook process

The typical use in a notebook process would have 4 points to take into account:

  • Have the environments created for each type of environment necessary in the notebooks

  • Using a %python interpreter other than “shared” to allow parallel launches of python processes

  • Activation at the start of the notebook of the desired environment to have the necessary libraries and versions

  • Disabling the environment when the notebook is terminated