I work on macOS Catalina but I imagine that this setup (mostly) works for earlier macOS versions.
In order to not mess up my system-installed Python distribution, I use pyenv to manage different Python versions. This is particularly helpful for me because I am working on multiple projects that usually use different Python interpreters at any given moment.
To install pyenv with Homebrew:
$ brew update
$ brew install pyenv
Once pyenv is installed, follow step 3 here.
To isolate projects from one another, I use poetry instead of managing my own virtualenv. Since poetry can detect whether I’m currently in an activated virtualenv, it will use that existing one or create a new one.
To install poetry:
$ curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | python -
When I use poetry, I either initialize a new project with poetry new or create a new project by hand with poetry init:
$ mkdir new-proj && cd new-proj
$ poetry init
All of my .virtualenvs also exist outside of my project folders and in my home directory ($HOME/.virtualenvs).
$ poetry config virtualenvs.path $HOME/.virtualenvs
The default .virtualenv path of poetry is {cache-dir}/virtualenvs which can be a bit harder to find depending on your OS. If you’re like me and you use PyCharm, finding that virtualenv so you can configure your Project Interpreter can be a pain.
I’ve found that this setup requires the least effort and works just as well when I need to work with scientific packages.