| layout |
|---|
reference |
In this course we have deliberately chosen to use a single approach to creating and activating Conda environments and installing packages within them, following the pattern of Create > Activate > Install. This was done to avoid confusing and overwhelming participants with the multitude of options because there are other methods/syntax for combining these steps that achieve the same end result.
It is possible to create a Conda environment and install specific Python packages within it (from Conda repositories
at least, not using pip to install packages from PyPI) in a single step. This negates the need to
specify a version of Python to install (i.e. python=3.10) as it will be pulled in as a dependency of the Python
package.
For example to create the basic-scipy-env created whilst working through Chapter 2 you could use the following.
$ conda create --name basic-scipy-env numba scikit-learn=1.2.0 ipython matplotlib=3.7 scipy=1.9.3
$ conda activate basic-scipy-env
{: .language-bash}
An alternative method of specifying the channel from which you wish to install a specific package from is know as "Namespaces" where the package is preceded by the channel from which you wish to install it from and separated by two semi-colons.
For example to install the polars package from the conda-forge channel within the my-final-project environment as
is done Chapter 3 you could use the following.
$ conda create --name my-final-project python=3.10
$ conda activate my-final-project
$ conda install jupyterlab matplotlib conda-forge::polars
{: .language-bash}
This can even be combined with the above example of creating and installing packages at the same time to give the following.
$ conda create --name my-final-project python=3.10 jupyterlab matplotlib conda-forge::polars
$ conda activate my-final-project
{: .language-bash}
{% include links.md %}