8 Using Conda Environment

Note: this page is moved from my ml-101 repo directly.

This doc is for Ubuntu/Linux only, for now.

One could install conda at anaconda.com. Once finished installation, you can edit .bashrc file in your home directory, and comment out “source” line that added by the Anaconda installer so that you won’t mess up anaconda python and OS python. See below section to check how to activate conda.

Similar with python virtual environment, conda has its own environment. The detailed documentation could be found conda.io.

For this project, I created an environment called coding-learning-env. You can create the same environment by using the “conda-env.yml” file.

8.0 How to create a conda env

The first thing first to learn is to create a conda environment. Creating a conda env is as easy as:

conda create --name my-conda-env

where the “my-conda-env” is the name of your environment. You can also specify the version of python you choose to use in that environment:

conda create -n my-conda-env python=3.6

In this case, we create the env by following command:

conda create -n coding-learning-env python=3.6

8.1 How to activate conda

1. Use command in terminal as follow: source anaconda3/bin/activate

   Please notice that you do not need to activate conda if the activation is already writen in your .bashrc file.

8.2 How to activate conda environment

1. Activate conda

2. Use command in terminal as follow: source activate coding-learning-env

8.3 How to create the same environment

When there is a yml file existed in the repo, and you want to create the same environment with same pacakges, you can follow steps below:

1. Activate conda

2. Git clone the repo to local, then cd to that repo directory.

3. Use command in terminal as follow: conda env create -f conda-env.yml

8.4 How to exit current environment

1. Use command in terminal as follow: source deactivate

   Plus, this works for both leaving conda and conda environment.

8.5 How to export the conda environment to yml file

1. Activate the conda environment you want to export first.

2. Use command in terminal as follow: conda env export > nameEnvironment.yml, where nameEnvironment is the yml file name.

   In this project case, we use conda env export > conda-env.yml

   Plus, this is how you can update the conda environment.

3. To obtain no-builds list, one could use conda env export --no-builds > conda-no-builds-env.yml

   The no-builds list is for cross-platform.

8.6 How to see the conda environment list

1. Activate conda

2. Use command in terminal as follow: conda env list