What is Anaconda?
Anaconda is both a data science platform and distribution of Python. But its also a nice way of managing multiple versions of Python and Virtual Environments (in the same way as VirtualEnv).
To install go to the Anaconda Downloads page and download the version dependent on the version of Python you would like. I used the graphical installer which is really straightforward.
Only Use Anaconda Python in an Anaconda Virtual Environment
Once the install is complete if you reload your terminal and check your python version (assuming you installed Python 3):
> python3 -V >> Python 3.4.5 :: Anaconda 4.1.1 (x86_64)
This is fine if you want to use the Anaconda variant of Python all the time. But what if you want to use vanilla Python when not in a Virtual Environment? Perhaps you have a tool (like I did) which does not play nice with Anaconda.
To fix this open up your bash_profile:
> open ~/.bash_profile
The file has been updated with a new statement adding Anaconda to the beginning of the path:
# added by Anaconda3 4.1.1 installer export PATH="/Users/thomasshipley/anaconda/bin:$PATH"
Update this so that Anaconda is added to the end of your path:
# added by Anaconda3 4.1.1 installer export PATH="$PATH:/Users/thomasshipley/anaconda/bin"
Restart your terminal and now the Anaconda variant of Python is only used when you are within an Anaconda Virtual Environment. To confirm this check your version of python the terminal:
> python3 -V >> Python 3.5.2
Creating and Activating a Virtual Environment
Now you have Anaconda installed it is easy to create a virtual environment to work in:
> conda create -n my_env python=3.4 anaconda
The command above is quite straightforward just pass an -n flag for the name of your environment and specify the python version in the python flag – in this case 3.4.
Activate and Deactivate Your Environment
To activate your new environment:
> source activate my_env
And to deactivate your environment:
> source deactivate
Now you can switch between an unlimited number of environments with specific python versions easily.