Starting the Command Line Interface

On Windows, launch the Anaconda Prompt via the Start menu; on macOS or Linux, open a terminal window.

The CLI opens with the base environment activated, created automatically during Anaconda installation, which includes a Python installation along with core libraries and dependencies for conda.

Here’s what the CLI looks like on Windows, where I’ve entered the command conda info, revealing important information such as package cache location and the active Python version.

In the CLI, the conda command serves as the primary interface for managing environments and package installations. You can use it to:

• Query and search the Anaconda package index and your current installation.
• Create and manage conda environments.
• Install and update packages in existing environments.

To begin, either create a new conda environment or activate an existing one. The following table outlines some useful single-line conda commands for managing environments. Replace the all-uppercase terms with specific names.

Note that many command options starting with two dashes can be abbreviated. For instance, use -n instead of --name.

Here’s an example command to create a new conda environment named my_first_env and install the latest Python version:

For a comprehensive list of commands, refer to the “conda cheat sheet.”

Ensure that Anaconda is installed correctly by following the appropriate instructions, which guarantees that Anaconda is added to your PATH, allowing conda commands to function in the terminal on macOS and Linux.

Creating a New Conda Environment

Let’s create a new conda environment called my_first_env. In the Anaconda Prompt or terminal, type:

conda create --name my_first_env python

Confirm by entering y when prompted. To bypass this verification prompt in future commands, append the --yes or -y flag, though it’s advisable to avoid this for everyday tasks to minimize errors.

If you wish to install a specific version of Python, such as 3.9, use:

conda create --name my_first_env python=3.9

This command installs the latest version from the Python 3.9 series. To obtain an exact version, use a double equal sign:

conda create --name my_first_env python==3.9.0

To install multiple packages while creating an environment, list them after the Python installation:

conda create --name my_first_env python numpy pandas

To activate the new environment, enter:

conda activate my_first_env

Next, confirm that the environment was created and is currently active by executing:

conda env list

This will display a list of environments, with the active one indicated by an asterisk (*).

The command prompt now reflects the active environment's name, providing clear visibility of which environment is currently in use.

To view the packages installed in the active environment, type:

conda list

To inspect a non-active environment, such as vote_counts, use:

conda list -n vote_counts

Remember that -n is shorthand for --name.

Specifying an Environment's Location

By default, conda environments are stored in the envs folder beneath your Anaconda installation. For instance, on my Windows setup, the environment we created is located at:

C:Usershannaanaconda3envsmy_first_env

However, you can store the environment elsewhere, allowing you to keep it within a project directory for easier access and sharing.

To create a conda environment outside the default location, replace the --name flag with --prefix:

conda create -p D:anywhere_you_wanta_projectconda_env

To activate this environment, run:

conda activate D:anywhere_you_wanta_projectconda_env

Although your new environment will appear in the output of conda env list, it won’t have a designated name. Here's an example where the non-default environment is marked by an asterisk:

There are some drawbacks to specifying a location other than the default. For instance, conda will not recognize your environment using the --name flag.

To list the contents of my_first_env in the default location, simply enter:

conda list -n my_first_env

For environments located elsewhere, use the --prefix flag:

conda list -p D:anywhere_you_wanta_projectconda_env

Another consequence is that the command prompt will display the active environment’s full path instead of its name, potentially resulting in lengthy prompts.

You can instruct conda to always display the environment's name in the prompt by modifying the env_prompt setting in the .condarc file, which is an optional configuration file for advanced users.

To adjust the prompt, use:

conda config --set env_prompt ?({name})?

This will ensure that only the environment name appears in the prompt, regardless of its storage location.

# Package Management

Once an environment is established, conda allows you to check available packages, find specific ones, install them, and update or remove packages. It’s advisable to install all required packages for a project simultaneously to prevent dependency conflicts.

The following table illustrates some practical conda commands for managing packages, primarily within active environments.

For a complete list, refer to the “conda cheat sheet.”

Installing Packages

The best practice for installing packages with conda is from within an active environment. Alternatively, you can install packages from outside an environment using the --name or --prefix flag with a directory path, though this approach is not recommended due to the risk of placing packages in the wrong environment.

To add two packages—Matplotlib and pillow—to my_first_env, first activate the environment:

conda activate my_first_env

It’s advisable to specify the version for each package during installation to ensure you capture the specific state of your environment for future sharing or rebuilding. To check for Matplotlib's current version, search using:

conda search matplotlib

The output will display a list of available versions, along with channel information.

The pkgs/main channel is the top priority in conda's defaults channel, which is set to the Anaconda Repository. Packages in conda-forge may be more current than those in defaults, but the defaults channel guarantees compatibility.

If no channel is specified, Anaconda will default to the highest-priority channel in your .condarc file. To view your channels, run:

conda config --show channels

If the defaults channel appears lower in priority than conda-forge, you can adjust the order.

To install both packages in my_first_env, specifying the latest versions:

conda install matplotlib=3.8.2 pillow=10.2.0

After installation, verify by entering:

conda list

To force conda to bypass channel priority, use:

conda config --set channel_priority false

You can also specify a channel using the --channel flag:

conda install -c defaults matplotlib=3.3.4

To adjust the order of channels in your configuration, use --remove, --append, or --prepend. Ideally, the defaults channel should be prioritized.

To automatically install a base package in every new environment, edit your configuration file:

conda config --add create_default_packages python

By doing this, Python will be included by default in every new conda environment.

To review your default packages list, use:

conda config --show

To remove a package from the default list, substitute --remove for --add. For more options, consult conda config --help.

Updating Packages

As new versions of installed packages become available over time, the following commands help keep your environment current.

First, ensure conda itself is up to date:

conda update -n base conda

To check for updates to a specific package, like pip, run:

conda update pip

If updates are available, you’ll see the new package information and be prompted to accept or decline the update.

To update all packages in an active environment to their latest versions, enter:

conda update --all

To update a non-active environment, use:

conda update -n <ENV_NAME> --all

If conflicting constraints exist within your environment, conda might resort to older package versions to satisfy dependencies.

Avoid managing specific package versions in the base environment; instead, utilize dedicated conda environments for precise package control.

For additional information on these topics, refer to the documentation.

Removing Packages

To remove a package, such as Matplotlib, from an active environment, enter:

conda remove matplotlib

To eliminate multiple packages simultaneously, list them:

conda remove matplotlib pillow

For a package in a non-active environment, specify the environment name:

conda remove -n <ENV_NAME> matplotlib

To verify the results of updates and removals, utilize:

conda list

# Duplicating and Sharing Environments

You can duplicate a conda environment by cloning it or using a file that lists its contents, facilitating sharing and archiving.

Cloning Environments

The simplest method to duplicate an environment is with the --clone flag:

conda create --name my_second_env --clone my_first_env

To confirm, run:

conda env list

Using an Environment File

You can also duplicate an environment by exporting its contents to an environment file, a YAML text file that lists all packages and versions installed within an environment.

To create an environment file for my_second_env, activate it and run:

conda activate my_second_env conda env export > environment.yml

You can name the file anything valid, but be cautious as existing files with the same name will be overwritten. By default, this file is saved in your user directory.

You can share this file with others, allowing them to replicate your environment. To create a file that works across platforms:

conda env export --from-history > environment.yml

This will include only explicitly requested packages, without their dependencies.

To re-create an environment from an environment.yml file:

conda env create -n my_second_env -f directorypathtoenvironment.yml

To update an existing environment using the file, run:

conda env update -n <ENV_NAME> -f directorypathtoenvironment.yml

For more information on environment files, consult the documentation.

Using a Specifications File

If your environment does not include pip-installed packages, you can create a specifications file for accurate duplication on the same operating system. To create this file, activate an environment and run:

conda list --explicit > exp_spec_list.txt

To re-create the environment using this text file:

conda create -n my_second_env -f directorypathtoexp_spec_list.txt

This specifications file will indicate the targeted platform.

Restoring Environments

Conda maintains a history of changes made to environments, allowing you to revert to prior versions. To view available versions, activate the environment and run:

conda list --revisions

Each revision is marked, and to restore to a previous version, use:

conda install --revision 3

If you roll back to an earlier revision, it will receive a new number.

Removing Environments

To delete a conda environment, first deactivate it:

conda deactivate

Then remove it with:

conda remove -n <ENV_NAME> --all

To verify removal, run:

conda env list

For environments outside of the default folder, include the directory path:

conda remove -p <PATHENV_NAME> --all

Cleaning the Package Cache

Over time, the anaconda3 folder can accumulate unnecessary files. To reclaim space, clean the package cache:

conda clean --all --dry-run

To proceed with cleaning:

conda clean --all

This command will remove index cache, unused packages, tarballs, and lock files. Windows users should reboot after executing this command.

For further options with conda clean, refer to the documentation.

# Conclusion

Every Anaconda Python project should utilize a conda environment to maintain organization, isolation, and reproducibility. These environments are essentially dedicated folders in your directory structure, and they can be stored together or alongside project folders.

While Anaconda Navigator offers a simple interface, leveraging command line interface commands is far more efficient for managing environments and packages.

# ADDENDUM: Working with PIP and Conda

Occasionally, you may encounter packages that cannot be installed using conda, necessitating the use of pip. Here are some guidelines for working with both systems:

Conda and pip are similar, but with two notable differences: pip exclusively manages Python packages, while conda handles multiple languages, and pip installs from the Python Package Index (PyPI) while conda sources from the Anaconda repository.

When using pip within a conda environment, follow these best practices:

• Install pip packages last.
• Avoid running pip in the root environment.
• If changes are necessary, recreate the conda environment.
• Maintain both conda and pip requirements in an environment file.

When using conda list, packages installed via pip will display with a “pypi” channel designation.

For further details, consult the blog for additional insights on managing both package managers.

# Thank You!

Thank you for reading! Follow me for more Quick Success Data Science projects in the future.

# Engaging with the Community

Thank you for being part of the In Plain English community! Before you go:

• Be sure to clap and follow the writer!
• Follow us on social media platforms.
• Explore our other offerings at Stackademic, CoFeed, and Venture.
• Discover more content at PlainEnglish.io.