4.4 Installing Packages

You can install packages from Conda’s repositories or from the conda-forge channel.
For packages not available through Conda, you can safely use pip inside a Conda environment.

Install a Single Package

conda install numpy

Install Multiple Packages

conda install numpy pandas matplotlib

Install a Specific Version of a Package

If you need a particular version of a package, specify it after the package name:

conda install numpy=1.26

You can also combine version specifications when installing multiple packages:

conda install pandas=2.2 matplotlib=3.8

---

Install from Conda-Forge

conda install -c conda-forge scikit-learn

---

Install pip Inside a Conda Environment

conda install pip

After this, you can use pip within that environment to install packages not available on Conda:

pip install some-package

💡 It’s recommended to use pip only inside an active Conda environment. This keeps your global Python setup clean and ensures all dependencies remain isolated.

---

Finding Package Names and Installation Commands

Sometimes the package name on Conda differs from its PyPI or GitHub name. To find the correct installation command:

• Search online using:

  conda search <package-name>

  Example:

  conda search tensorflow

• Or use:

  • Anaconda package search
  • Conda-Forge package search

These resources show exact installation commands, available versions, and the correct channels (for example, -c conda-forge).

---