Conda

Fatma Kahveci
Coding
Page content
  • It is a package, dependency, and environment management.
  • It was created for Python, but it is also used for R, Java, C/C++, etc.
  • It runs on macOS, Linux, etc.
  • If you need a package that requires another version of Python, use conda.
  • Use the most recent version.

miniconda vs. anaconda

miniconda

  • Choose if you are a newbie to python and have the time and disk space.

  • Install 1,500 scientific packages automatically installed at once.

  • Every time you open up a terminal, conda won’t automatically be available. Run the command below to use conda within miniconda.

export PATH=$HOME/miniconda/bin:$PATH

anaconda

  • Install each package separately.

Installation

Update packages

> conda update conda

  • It is friendly with pip. It means that if you can’t find a package in Anaconda Cloud, you can install it using pip.

Creating a conda environment

  • There are two ways of creating a conda environment.

1. An environment file in YAML

  • YAML := YAML Ain’t Markup Language

environment.yml

name: <environment_name>
channels:
- conda-forge
dependencies:
- python=<python_version>
- pip
- pip:
    - <pip_package_name>
  • To create an environment:

> conda env create --file <environment.yml>

2. Manual specification of packages

> conda create -c conda-forge -n <environment_name> python=<python_version> <pip_package_name_1> <pip_package_name_2> <...>

  • conda env remove -n live_env

  • To activate an environment:

> conda activate <environment_name>

  • To deactivate an environment:

> conda deactivate

  • To delete an environment:

> conda env remove -n <environment_name>

  • To see the packages in an environment:

> conda list -n test_env

  • To add a package to an existing environment:

> conda install --name <environment_name> -c <chanel_name> <package_name>

  • Get a list of all my environments (Active environment shown with *)

> conda info -e

References