Linux is GREAT !!! Windows is GREAT !!! Which one ???
I am not contesting for any of the opponents above. Personally what ever does the job in the correct and suited manner for the business is a winner.
In this post I am trying to bring together the facts, which motivated our decision to opt for Linux (any distro, opted for ubuntu) as our python development platform.
Recently when I started my journey in the Python universe, I realised that (having a strong background in windows development), I setup the development environment in Windows for my data science projects.
Which means I got started, installing the following:
And started working in Jupyter Notebook, involving extracting data from SQL, analysing, statistics and progressing in the arenas of forecasting (time series).
As the journey continued within the space, I started encountering few hiccups (slight discomforts) when I encountered the following information on the few web sites:
Prophet – Facebook Open Source (depends on PyStan)
Then I was running the Jupyter notebook server instance from anaconda distribution, I felt it was a multi-user system which if installed on a server it would allow other members of the team to collaborate.
Then the following information surfaced:
Jupyter – Running a notebook server
So embarking to install a multi-user Jupyter server we came across the following information:
And when working with large data sets/series to forecast upon, multi-threading is the most important concepts to have in place.
And coming to this information directly from Python gave me enough evidence that Linux distributions are more suited for python based data science projects.
Multiprocessing — Process-based parallelism
I hope the above helps in an assimilated way for planning your development environment setup.