Setting Up Anaconda

Anaconda Distribution bundles Conda, Python, and many scientific packages. Miniconda provides Conda with a smaller base environment. Choose one from the official Anaconda installation guides and follow the instructions for your operating system and processor.

Verify Conda

Open Anaconda Prompt on Windows or a shell initialized for Conda on macOS/Linux:

conda --version
conda info

Do not add arbitrary Anaconda directories to PATH by hand. Use the installer and conda init guidance for your shell, then open a new shell if instructed.

Create a project environment

conda create --name analysis python=3.15
conda activate analysis
python --version
conda list

Use a Python version supported by the packages you need; the current Python release may not yet be available across every channel. Keep project packages out of the base environment.

Install packages from deliberately chosen channels:

conda install numpy pandas
conda list

Mixing channels can change dependency solutions. Record the channels and their priority, and do not mix pip and Conda casually. If a package is only available through pip, install Conda dependencies first, use the environment's python -m pip, and retest the complete environment.

Export and recreate

For a cross-platform history of explicitly requested packages, current Conda provides the enhanced conda export command:

conda export --from-history --format=environment-yaml --file=environment.yml
conda create --file environment.yml

--from-history excludes transitive platform-specific dependencies, so it is portable but not an exact lock. Conda also supports platform-specific explicit exports and multi-platform lock files for stricter reproduction. Inspect every exported file before sharing it because environment variables or channel URLs may contain sensitive information.

Leave or remove an environment with:

conda deactivate
conda env remove --name analysis